# Chain188 > Exchange data, independently checked > > Structured, answer-first crypto affiliate research published at https://www.caijing188.com/. Every page carries schema.org structured data (Organization, Article/Product+Offer, FAQPage, BreadcrumbList) and an explicit summary block that can be quoted verbatim. ## Publisher entity - Name: Chain188 - Also known as: Caijing 188 - Description: Independent data on crypto exchanges, wallets and crypto cards: volume, fees, access rules and the date each figure was checked. - Topics: stablecoins, security, liquidity, wallets, key management, liquidity risk, self-custody, blind signing, seed backup, phishing, hardware wallets, rollups ## Live affiliate offers (updated 2026-10-27T23:35:51+08:00) - [Binance welcome package: up to $600 in vouchers](https://www.caijing188.com/deals/binance-welcome-package/): Up to $600 · code GRO_28502_L6OOC · expires 2026-11-17 - [OKX mystery box: up to $10,000 after your first deposit](https://www.caijing188.com/deals/okx-mystery-box/): Up to $10,000 · code 2BDDHB6F · expires 2026-11-24 - [KuCoin welcome reward: up to $700 for new accounts](https://www.caijing188.com/deals/kucoin-welcome-reward/): Up to $700 · code KUCOIN700 · expires 2026-12-08 - [Bybit demo trading: $30,000 practice balance](https://www.caijing188.com/deals/bybit-demo-bonus/): $30,000 demo · code BYBITDEMO · expires 2026-12-01 ## Exchange & platform entities - [Binance](https://www.caijing188.com/exchange/binance/): Exchange · taker 0.1% · rated 4.8/5 · bonus: Up to $600 welcome package · available: Global (excl. US for some products) - [OKX](https://www.caijing188.com/exchange/okx/): Exchange · taker 0.08% · rated 4.6/5 · bonus: Mystery box up to $10,000 · available: Global (excl. US for some products) - [Bybit](https://www.caijing188.com/exchange/bybit/): Exchange · taker 0.1% · rated 4.5/5 · bonus: $30,000 demo trading bonus · available: Global (excl. US for some products) - [Kraken](https://www.caijing188.com/exchange/kraken/): Exchange · taker 0.26% · rated 4.4/5 · available: Global (excl. US for some products) - [Coinbase](https://www.caijing188.com/exchange/coinbase/): Exchange · taker 0.6% · rated 4.3/5 · bonus: $10 in BTC after first trade · available: Global (excl. US for some products) - [Bitget](https://www.caijing188.com/exchange/bitget/): Exchange · taker 0.1% · rated 4.2/5 · bonus: Up to $6,200 welcome pack · available: Global (excl. US for some products) - [Gate.io](https://www.caijing188.com/exchange/gate-io/): Exchange · taker 0.15% · rated 4.1/5 · bonus: Up to $6,666 bonus · available: Global (excl. US for some products) - [Uniswap](https://www.caijing188.com/exchange/uniswap/): Exchange · taker 0.3% · rated 4.1/5 · available: Global - [Crypto.com](https://www.caijing188.com/exchange/crypto-com/): Exchange · taker 0.075% · rated 4.0/5 · bonus: $50 in CRO after staking · available: Global (excl. US for some products) - [KuCoin](https://www.caijing188.com/exchange/kucoin/): Exchange · taker 0.1% · rated 4.0/5 · bonus: Up to $700 welcome reward · available: Global (excl. US for some products) - [1inch](https://www.caijing188.com/exchange/1inch/): Exchange · rated 3.9/5 · available: Global - [Upbit](https://www.caijing188.com/exchange/upbit/): Exchange · taker 0.25% · rated 3.9/5 · available: Global (excl. US for some products) - [Bitstamp](https://www.caijing188.com/exchange/bitstamp/): Exchange · taker 0.5% · rated 3.8/5 · available: Global (excl. US for some products) - [CoW Swap](https://www.caijing188.com/exchange/cowswap/): Exchange · rated 3.8/5 · available: Global - [Gemini](https://www.caijing188.com/exchange/gemini/): Exchange · taker 0.4% · rated 3.7/5 · available: Global (excl. US for some products) - [PancakeSwap](https://www.caijing188.com/exchange/pancake-swap/): Exchange · taker 0.25% · rated 3.7/5 · available: Global - [Balancer](https://www.caijing188.com/exchange/balancer/): Exchange · taker 0.1% · rated 3.6/5 · available: Global - [Bithumb](https://www.caijing188.com/exchange/bithumb/): Exchange · taker 0.25% · rated 3.6/5 · available: Global (excl. US for some products) - [CoinEx](https://www.caijing188.com/exchange/coinex/): Exchange · taker 0.2% · rated 3.5/5 · available: Global - [SwissBorg](https://www.caijing188.com/exchange/swissborg/): Exchange · taker 0.25% · rated 3.4/5 · available: Global (excl. US for some products) - [Paribu](https://www.caijing188.com/exchange/paribu/): Exchange · taker 0.2% · rated 3.2/5 · available: Global (excl. US for some products) - [BTCTurk](https://www.caijing188.com/exchange/btcturk/): Exchange · taker 0.2% · rated 3.1/5 · available: Global (excl. US for some products) - [SushiSwap](https://www.caijing188.com/exchange/sushiswap/): Exchange · taker 0.3% · rated 3.1/5 · available: Global - [Bancor](https://www.caijing188.com/exchange/bancor/): Exchange · taker 0.1% · rated 3.0/5 · available: Global ## Guides and reviews (with answer summaries) - [The fifteen-day exit: what the ETH queue costs a fund](https://www.caijing188.com/fifteen-day-exit/): The ETH exit queue held 850,000 ETH in October 2026, about 14.8 days of waiting, against roughly 2.5 million ETH trying to enter on 22 July 2026. For a fund, 14.8 days of illiquidity is worth about 0.11% of gross staking yield, as of October 2026. - [Impermanent loss, worked through with one pool and one price move](https://www.caijing188.com/impermanent-loss-one-pool/): Impermanent loss is the gap between a pool position and simply holding, not a loss against your deposit. In a 50/50 pool where one asset doubles, the position ends 5.72% below holding, as of October 2026. Fees must beat that gap, and the quoted fee APR is trailing rather than forward. - [What a listing announcement actually moves, and how fast](https://www.caijing188.com/listing-announcement-liquidity/): A listing announcement moves price in minutes and book depth over days, so day-one volume mostly reflects market makers and bots, as of October 2026. Our board ranks 24 venues, and FeeEdge's 17 August 2026 survey put entry-tier spot taker fees near 0.10%. Check depth at your size before you size. - [MiCA passporting: which entity actually holds your account](https://www.caijing188.com/mica-entity-passporting/): A passporting licence lets one authorised firm serve several markets, so the contracting entity in your terms can differ from the brand on the homepage, as of October 2026. Our board ranks 24 venues, and reserve disclosures remain snapshots rather than guarantees. Read the client agreement, the register entry and the reserve page, in that order. - [Blob space after EIP-4844: what an L2 does when data gets expensive](https://www.caijing188.com/blob-space-pricing/): Blob space decoupled L2 data cost from mainnet gas, but it is its own market with its own price. Ethereum.org's snapshot, as of October 2026, listed Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction. When the lane is contested a rollup either pays more or posts less often, and both reach the user. - [Hardware wallets: what the secure element does not protect](https://www.caijing188.com/secure-element-limits/): The secure element keeps the key inside the chip. It does not read the transaction for you. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026, moved by signatures. If the device shows a hash instead of an amount and a destination, the chip is protecting a key you are spending blind. - [Revenue per transaction is the L2 metric that survives incentives](https://www.caijing188.com/revenue-per-transaction/): Base charged $0.001 to $0.002 per transfer and Arbitrum One $0.004 to $0.005 in Ethereum.org's 1 to 6 October 2026 snapshot, as of October 2026. Revenue per transaction turns a headline into a transfer count: a million dollars is 500 million to 1 billion transfers on Base and 200 to 250 million on Arbitrum. - [Liquid staking tokens: the depeg nobody models](https://www.caijing188.com/lst-depeg-risk/): A liquid staking token converts the validator exit queue into a secondary-market price. Ethereum's exit queue held 850,000 ETH at about 14.8 days in October 2026, while Lido's 10% fee on a 2.63% gross leaves 2.37%, as of October 2026. Price the worse of the two exits. - [Order book depth is the number your slippage model is missing](https://www.caijing188.com/depth-over-volume/): 24h volume measures turnover, not resting size, and a venue can rank high on our board of 24 venues as of October 2026 while staying thin inside one percent. FeeEdge's 17 August 2026 survey puts Solana withdrawals near 0.5 to 1 USDT. Read depth at your size minutes before the order, not the day before. - [Stablecoin rules did not kill yield, they moved it](https://www.caijing188.com/stablecoin-yield-rules/): Rules that stop a stablecoin issuer paying a return push the return into a separate product, as of October 2026, when tokenised Treasury bills yielded 3.60% to 3.61% on a seven-day basis (RWA.xyz, 7 October 2026). Total stablecoin supply was $292.40B that month. Check what you hold, not what the page calls it. - [Finality is a budget, not a switch](https://www.caijing188.com/finality-as-a-budget/): Finality is a trade of time for certainty, not a binary state. Ethereum.org's directory, as of October 2026, lists 7-day exits on Base and OP Mainnet, a 45,818-block dispute window on Arbitrum, about 6.4 days, and a 3-hour minimum on ZKsync Era. An exchange credits on a confirmation count; a bridge waits for the window. - [Seed backups that survive a fire, and the one that does not](https://www.caijing188.com/seed-backup-that-survives/): A seed in one location loses to fire, water or burglary, and a photograph loses to camera roll sync. The stablecoin float was $292.40B as of October 2026, per this board's data pack, and much of it sits behind one piece of paper. Split the backup across locations, keep it offline, and test the restore while the balance is small. - [When a stablecoin’s $292B float is really three balance sheets](https://www.caijing188.com/float-concentration/): Stablecoin supply reached $292.40 billion in October 2026, but that is a stock, not a measure of who holds it. Issuer share and chain share describe rules and routes, as of October 2026. Only holder concentration predicts how fast a stressed float can exit. - [Collateral you cannot liquidate is not collateral](https://www.caijing188.com/illiquid-collateral/): Tokenised Treasuries carry almost no price risk and a great deal of time risk, because most redeem only through the issuer in business hours after KYC. DefiLlama put on-chain AUM at $34.865 billion against $4.182 billion deployed in DeFi in October 2026, as of October 2026. The gap is liquidity, not demand. - [Why two exchange rankings disagree about the same venue](https://www.caijing188.com/volume-rankings-disagree/): Two rankings can disagree about one venue because pair universe, wash filters, quote currencies and sampling windows differ. Our board ranks 24 venues on sampled 24h volume as of October 2026, with the top row near a third of the sample. Read the methodology before quoting a rank. - [Staking rewards: income at receipt, gain on sale](https://www.caijing188.com/staking-rewards-tax/): The value of a staking reward when it lands becomes your cost basis, so the timestamp on the reward record matters more than the annual rate, as of October 2026. ETH staking paid 2.63% gross in Q3 2026, and over 96% of it came from issuance (Bitwise, Q4 2026). This is not tax advice. - [Bridges are three products, and only one of them is canonical](https://www.caijing188.com/three-kinds-of-bridges/): Three products share the word bridge: the rollup's canonical bridge, a third-party liquidity network, and a wrapped-asset issuer. Each holds keys differently and pays you differently. Ethereum.org listed, as of October 2026, 7-day canonical exits on Base and OP Mainnet and a 45,818-block window on Arbitrum. Ask what claim you hold before you move size. - [Multisig for a small team: the setup that does not lock you out](https://www.caijing188.com/multisig-small-team/): A 2-of-3 multisig survives one lost key and one compromise, which is why it fits a small team, while 3-of-5 suits a treasury and needs five maintained signers. Staked ETH showed the availability side as of October 2026, with an exit queue of 850,000 ETH, about 14.8 days. Test a recovery before funding. - [What it costs to move $20 on eight networks](https://www.caijing188.com/cost-of-twenty-dollars/): Moving $20 cost under a cent on Solana and $2 to $10 on Ethereum L1 ERC-20 in FeeEdge's 17 August 2026 survey, or 0.01% to 50% of the amount, as of October 2026. Tron's $1 to $3 works out at 5% to 15% of the same transfer. Batch small moves. - [Reading a vault APY: base yield, incentives, and the part that expires](https://www.caijing188.com/vault-apy-anatomy/): A vault APY usually adds three legs: base yield, an incentive token, and points. Only the base is a rate, and only the net after fees and slippage compounds. Maple paid 5.20% and Plume nOPAL 10.73% against 3.60% to 3.61% on tokenised Treasuries, as of October 2026, which is credit risk rather than a risk-free rate. - [The same coin costs more in some markets, and it is not arbitrage](https://www.caijing188.com/regional-premium-pricing/): A local premium is a friction price, set by banking rails, capital controls and withdrawal windows rather than by the coin. Stablecoin supply reached $292.40B as of October 2026 per RWA.xyz. Capturing the gap needs a local bank account, so most retail traders outside the market cannot. - [KYC tiers change your withdrawal limit more than your privacy](https://www.caijing188.com/kyc-tiers-withdrawal-limits/): Verification tiers gate withdrawal limits and payment rails, not just identity, as of October 2026. Our board tracks 24 venues, and a Solana SPL withdrawal costs about 0.5 to 1 USDT while an ERC-20 withdrawal costs $2 to $10 in FeeEdge's 17 August 2026 survey. Read the tier table before you fund. - [MEV on rollups: the sequencer sees your transaction first](https://www.caijing188.com/mev-on-rollups/): A single sequencer holds the ordering right, so a rollup swap is not a public mempool race but a private queue with an operator's policy attached. Ethereum.org's snapshot, as of October 2026, listed Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction. Ask what mechanism backs a protection claim before trusting the label. - [Approval hygiene: the monthly revoke that saves you](https://www.caijing188.com/approval-hygiene/): An ERC-20 approval outlives the site that requested it, and an unlimited allowance drains without a second signature. Revoking costs what a transaction costs: $2 to $10 on Ethereum L1 in FeeEdge's 17 August 2026 sample, and $0.001 to $0.002 on Base as of October 2026. Review monthly from a spending wallet. - [Oracle latency is a liquidation risk, not a data problem](https://www.caijing188.com/oracle-latency-liquidations/): A lending market liquidates against the last oracle write, not the market price, so a stale feed can make a safe position callable. The deviation threshold matters more than the update interval, and the buffer to liquidation is thin by design. Read the feed configuration, not the market page, as of October 2026. - [Stablecoin depeg: what actually happens in the first hour](https://www.caijing188.com/stablecoin-depeg-first-hour/): In the first hour of a depeg, redemption queues and thin books matter more than the last print. Stablecoin supply was $292.40B as of October 2026 per RWA.xyz, and Base or OP Mainnet exits take seven days per Ethereum.org's 1 to 6 October 2026 snapshot. Check chain and venue before stress. - [The travel rule: what actually travels with your transfer](https://www.caijing188.com/travel-rule-in-practice/): Travel rule messages carry originator and beneficiary data between regulated venues, which is why a self-custody withdrawal can trigger extra questions, as of October 2026. Our board ranks 24 venues, and an ERC-20 withdrawal costs $2 to $10 (FeeEdge, 17 August 2026). A not supported message usually describes the sending venue's own list. - [What a node actually verifies for you](https://www.caijing188.com/what-a-node-verifies/): A light client, a full node and a hosted RPC endpoint answer different questions, and only the first two check the chain. Ethereum.org listed, as of October 2026, Base fees at $0.001 to $0.002 and Linea near $0.024, low enough that most users never question the endpoint. Balances and nonces can be verified; old history usually cannot. - [Blind signing: why the screen is the only signature you can read](https://www.caijing188.com/blind-signing-screen/): When a wallet cannot decode a payload it shows a hash and asks for approval, and a hash cannot be checked against intent. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026. The fix is a screen that renders the amount and the full destination, plus a policy of refusing anything else. - [Looping: where the leverage hides inside a four percent spread](https://www.caijing188.com/looping-hidden-leverage/): Looping turns a four point lending spread into a leveraged position: supply, borrow, supply again. Three steps on an 80% loan-to-value gives 2.44 times leverage and a health factor of 1.40, leaving about 28.5% of collateral fall before liquidation, as of October 2026. The spread sizes the trade; the health factor decides whether you keep it. - [Tokenised funds: the licensing question product pages skip](https://www.caijing188.com/tokenised-fund-licensing/): A token can be a fund unit, a debt claim or a receipt, and that label decides who may sell it and to whom, as of October 2026. Tokenised Treasury products held $14.64B to $14.90B across 117 funds (RWA.xyz), and most require KYC to mint or redeem. Check the label before the yield. - [Passphrase, second seed, or split: three ways to survive a loss](https://www.caijing188.com/second-factor-seeds/): A BIP39 passphrase defeats a thief who finds the seed, but a forgotten passphrase is total loss, a second seed doubles the bookkeeping, and splitting trades convenience for recovery risk. The stablecoin float was $292.40B as of October 2026, per this board's data pack. Match the scheme to the event you fear. - [Restaking points: what you are actually underwriting](https://www.caijing188.com/restaking-points-risk/): Restaking points are an unpriced claim on a token that may not exist, not a yield. The real term sheet is the slashing condition attached to each additional service, and the delegation bounds that decide whether your exposure can change without consent. Ethereum's exit queue held 850,000 ETH at about 14.8 days, as of October 2026. - [Why a clean wallet can still be frozen](https://www.caijing188.com/sanctions-screening-freeze/): Screening scores the cluster of addresses around yours, so an inbound transfer from a counterparty two hops back can taint a wallet you never used badly, as of October 2026. Our board ranks 24 venues. Keep the trail and use a labelled withdrawal address, so an appeal has documents to work with. - [Wallet drains start with a signature, not a hack](https://www.caijing188.com/drainers-start-with-a-signature/): Most drains are a signature on a permit or a fake mint page, not a broken curve. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026, and an approval can move a balance without a key being stolen. Simulate, read the request, and keep savings behind a key that never connects. - [Stablecoin supply hit $292B, and the holders are narrower than the number suggests](https://www.caijing188.com/stablecoin-supply-who-holds-it/): Stablecoin supply stood at $292.40B in October 2026, but the float that actually trades is concentrated in a handful of issuers and a handful of chains. A large supply number is not the same as deep liquidity. - [Why L2 fee revenue replaced TVL as the number traders actually watch](https://www.caijing188.com/l2-fee-revenue-not-tvl/): Fee revenue is a better activity signal than total value locked because it measures paying users rather than parked capital. Base and Arbitrum dominate both, but the gap between them is much wider on revenue than on TVL. - [Proof-of-reserves reports: the three numbers that actually matter](https://www.caijing188.com/proof-of-reserves-three-numbers/): A proof-of-reserves report is only useful if it states a coverage ratio above 100%, a timestamp you can compare against a known balance, and the exact set of entities and chains in scope. Without all three, it is a snapshot of unknown completeness. - [The basis trade is back, and it is why your funding rate looks wrong](https://www.caijing188.com/basis-trade-funding-rates/): A negative funding print is not automatically bearish. When the cash-and-carry trade is crowded, perpetual funding compresses toward zero or below even while spot demand is healthy, because the arbitrage sells the perp to hedge a spot long. - [What a sequencer actually does, and what breaks when it stops](https://www.caijing188.com/what-a-sequencer-does/): A sequencer orders transactions and posts them to the base chain. When it stops, users cannot get included, but the state already settled is not lost. What you lose is time to exit, which is exactly what matters in a stress event. ## Topic hubs - [Markets](https://www.caijing188.com/category/markets/): Volume, flows and the numbers behind the moves. - [Technology](https://www.caijing188.com/category/technology/): Chains, rollups, sequencing and the plumbing underneath. - [Regulation](https://www.caijing188.com/category/regulation/): Licensing, KYC, tax and what the rules actually require. - [DeFi](https://www.caijing188.com/category/defi/): Lending, liquidity and the risk inside the yield. - [Analysis](https://www.caijing188.com/category/analysis/): Long reads on structure, incentives and market plumbing. - [Wallets](https://www.caijing188.com/category/wallets/): Custody, signing habits and self-custody trade-offs. ## Machine readable endpoints - Sitemap: https://www.caijing188.com/wp-sitemap.xml - Deals archive: https://www.caijing188.com/deals/ - Brands archive: https://www.caijing188.com/exchange/ - Main RSS: https://www.caijing188.com/feed/ - Full-text digest: https://www.caijing188.com/llms-full.txt - Per-post plain text: https://www.caijing188.com/?crx_summary={post_id} --- # Full content digest > Plain-text rendering of every published article. Headings are marked with #, lists with -, tables are rendered as rows. ## The fifteen-day exit: what the ETH queue costs a fund URL: https://www.caijing188.com/fifteen-day-exit/ Updated: 2026-10-27T22:09:39+08:00 Verified: 2026-10-27 ### Quick answer The ETH exit queue held 850,000 ETH in October 2026, about 14.8 days of waiting, against roughly 2.5 million ETH trying to enter on 22 July 2026. For a fund, 14.8 days of illiquidity is worth about 0.11% of gross staking yield, as of October 2026. ### Key takeaways - 850,000 ETH queued to exit in October 2026, about 14.8 days - Roughly 2.5 million ETH was queued to enter on 22 July 2026, a wait above 43 days - 14.8 days of staking at 2.63% gross is worth about 0.11% of principal - Tokenised T-bills paid 3.60% to 3.61% on 7 October 2026, above the 2.63% gross staking print ![A withdrawal queue stretching back from a vault door](https://img1.ladyww.cn/picture/Picture00441.jpg) 850,000 ETH is queued to leave Ethereum staking in October 2026, and at the rate the queue is clearing that is about 14.8 days. The gross staking yield for Q3 2026 was 2.63%. For a fund that has to meet redemptions those two numbers belong in the same sentence, because one is a duration and the other is the price paid for accepting it. ## Two queues, one churn limit The queue has reversed. On 22 July 2026 roughly 2.5 million ETH was waiting to enter, a wait of more than 43 days. By October 2026 the exit queue held 850,000 ETH at about 14.8 days. Divide the two prints and the clearing rate is almost identical. The July queue implies roughly 58,000 ETH a day moving through the churn limit. The October queue implies roughly 57,000. Entry and exit draw on the same throughput budget, so the wait is set by queue length rather than by direction. That is the part worth internalising. The wait is not a protocol constant. It is a function of how many people are ahead of you. With 43.8 million ETH staked on 23 October 2026, the 850,000 ETH exit queue holds about 1.9% of all staked ETH. ## What 14.8 days of staking actually pays At 2.63% gross, one day of staking pays about 0.0072% of principal. Fourteen and eight tenths of a day pays about 0.11%. Net of the operator commissions in the table below, the same window is worth 0.08% to 0.09%. That is the entire carry for waiting two weeks. An entering validator earns it. An exiting one does not, because the wait is time outside the active set, earning nothing while remaining exposed to the price. Bitwise’s Q4 2026 work attributes more than 96% of ETH staking rewards to new issuance rather than to fees. The yield is a dilution transfer paid in the asset being diluted, which is why [staking yield is not income](/staking-yield-is-not-income/) in the sense a fixed-income desk means. It is a claim on a growing supply. ## The alternative settles in a block Tokenised Treasury bills yielded 3.60% to 3.61% annualised on a seven-day basis as of 7 October 2026. That is above the 2.63% gross staking print and well above any staking figure after a commission. The honest comparison is not the yield. It is the settlement. A token transfer settles in a block. Minting and redeeming most of these products requires KYC, so the atomic leg is the secondary transfer, not the primary issuance. A fund still goes through a dealer to get in and out of size. But the secondary leg is the one that matters at 3am. Distributed value across tokenised products reached $39.01 billion by 16 October 2026, up from about $21.5 billion at the start of the year. Tokenised Treasuries stood at $14.64 billion to $14.90 billion across 117 funds. [A handful of products carry most of that float](/thirty-nine-billion-three-products/), and duration is part of the reason the category grew. | Instrument | Headline yield | Time to cash | Source and date | |—|—|—|—| | ETH staking, gross | 2.63% | About 14.8 days | Q3 2026 print | | ETH staking, net | 1.95% to 2.1% | About 14.8 days | Q3 2026, after 35% to 39.95% commissions | | Tokenised T-bills | 3.60% to 3.61% | One block on transfer, KYC on mint | 7 October 2026, seven-day | | Staked ETH outstanding | 43.8 million ETH | Queue at 1.9% of stake | 23 October 2026 | ## Sizing the buffer, not the yield A 14.8-day exit is a liquidity requirement, and it has to be funded outside the position. For every 1% of NAV that can be redeemed inside a week, the fund needs 1% of NAV in spot ETH or stablecoins that are not in the queue. That cash earns nothing, or earns a stablecoin rate, and it dilutes the staking yield across the whole book. Stake 100% of the ETH sleeve and the buffer has to come from somewhere else. Stake 90% and hold 10% liquid, and a 2.63% gross yield on the staked part becomes about 2.37% on the sleeve before commissions. After a 35% operator commission, 2.37% is nearer 1.5%. The queue and the cash buffer together have taken more of the yield than the operator did. Solo validation has the same shape with a higher bar. It requires 32 ETH per validator, and the exit still goes through the same queue as everyone else’s. ## Verdict Duration is the risk and the yield is the distraction. Fourteen and eight tenths of a day of queue turns a 2.63% gross print into roughly 0.11% of carry per turn, and the position is illiquid for the whole window. A fund that can tolerate the wait should compare its net staking figure against a 3.60% Treasury token and count days, not basis points. A fund that cannot should size the liquid buffer first and treat the staking yield as what is left over. The queue is the product. The yield is the fee for standing in it. ### FAQ Q: Why does the ETH exit queue take 14.8 days? A: Validators enter and exit through a churn limit that caps how much stake can move per epoch, so the wait is a function of how much is queued. In October 2026 the exit queue held 850,000 ETH, which cleared at about 57,000 ETH a day and produced roughly 14.8 days. On 22 July 2026 the same mechanism held an entry queue of about 2.5 million ETH, a wait of more than 43 days. Q: Is 2.63% staking yield worth a two-week lock? A: On the numbers, the carry is about 0.11% of principal for 14.8 days at the 2.63% gross Q3 2026 print, and 0.08% to 0.11% after commissions. Whether that is worth it depends on the fund's redemption profile rather than on the yield itself. If 1% of NAV can be redeemed inside a week, the buffer that covers it has to be held outside the queue. Q: How does a tokenised Treasury bill compare with staked ETH? A: Tokenised T-bills yielded 3.60% to 3.61% annualised on a seven-day basis on 7 October 2026, above the 2.63% gross staking print. The larger difference is settlement: a token transfer settles in a block, while minting and redeeming most products requires KYC. Staked ETH pays less and takes about 14.8 days to reach cash. ## SushiSwap URL: https://www.caijing188.com/exchange/sushiswap/ Updated: 2026-10-27T20:02:29+08:00 Verified: 2026-10-27 ### Quick answer SushiSwap is a exchange venue with $267,000 in reported 24-hour volume and 127 markets as of October 2026. Spot trading costs 0.3% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $267,000 reported 24-hour volume - 127 listed markets - 0.3% taker fee, entry tier - Identity verification: No ### What SushiSwap is SushiSwap is a centralised exchange venue launched in 2020. At our last check it reported $267,000 of 24-hour turnover across 127 markets, with $29M in 30-day volume. ### Fees Spot trading is priced at 0.3% maker and 0.3% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. SushiSwap supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does SushiSwap charge per trade? A: 0.3% maker and 0.3% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does SushiSwap require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## Bancor URL: https://www.caijing188.com/exchange/bancor/ Updated: 2026-10-27T20:02:28+08:00 Verified: 2026-10-27 ### Quick answer Bancor is a exchange venue with $2.1M in reported 24-hour volume and 68 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $2.1M reported 24-hour volume - 68 listed markets - 0.1% taker fee, entry tier - Identity verification: No ### What Bancor is Bancor is a centralised exchange venue launched in 2017. At our last check it reported $2.1M of 24-hour turnover across 68 markets, with $120M in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. Bancor supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Bancor charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Bancor require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## PancakeSwap URL: https://www.caijing188.com/exchange/pancake-swap/ Updated: 2026-10-27T20:02:28+08:00 Verified: 2026-10-27 ### Quick answer PancakeSwap is a exchange venue with $9M in reported 24-hour volume and 320 markets as of October 2026. Spot trading costs 0.25% taker on the entry tier. Identity verification is no. Headquartered in Decentralised (BNB Chain). ### Key takeaways - $9M reported 24-hour volume - 320 listed markets - 0.25% taker fee, entry tier - Identity verification: No ### What PancakeSwap is PancakeSwap is a centralised exchange venue launched in 2020. At our last check it reported $9M of 24-hour turnover across 320 markets, with $476M in 30-day volume. ### Fees Spot trading is priced at 0.25% maker and 0.25% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised (BNB Chain). Identity verification: No. Available in: Crypto-only, no identity check on the trading account. PancakeSwap supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does PancakeSwap charge per trade? A: 0.25% maker and 0.25% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does PancakeSwap require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## Balancer URL: https://www.caijing188.com/exchange/balancer/ Updated: 2026-10-27T20:02:27+08:00 Verified: 2026-10-27 ### Quick answer Balancer is a exchange venue with $14M in reported 24-hour volume and 210 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $14M reported 24-hour volume - 210 listed markets - 0.1% taker fee, entry tier - Identity verification: No ### What Balancer is Balancer is a centralised exchange venue launched in 2020. At our last check it reported $14M of 24-hour turnover across 210 markets, with $780M in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. Balancer supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Balancer charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Balancer require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## CoW Swap URL: https://www.caijing188.com/exchange/cowswap/ Updated: 2026-10-27T20:02:27+08:00 Verified: 2026-10-27 ### Quick answer CoW Swap is a exchange venue with $18M in reported 24-hour volume and 180 markets as of October 2026. Spot trading costs 0% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $18M reported 24-hour volume - 180 listed markets - 0% taker fee, entry tier - Identity verification: No ### What CoW Swap is CoW Swap is a centralised exchange venue launched in 2022. At our last check it reported $18M of 24-hour turnover across 180 markets, with $920M in 30-day volume. ### Fees Spot trading is priced at 0% maker and 0% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. CoW Swap supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does CoW Swap charge per trade? A: 0% maker and 0% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does CoW Swap require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## 1inch URL: https://www.caijing188.com/exchange/1inch/ Updated: 2026-10-27T20:02:26+08:00 Verified: 2026-10-27 ### Quick answer 1inch is a exchange venue with $31M in reported 24-hour volume and 240 markets as of October 2026. Spot trading costs 0% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $31M reported 24-hour volume - 240 listed markets - 0% taker fee, entry tier - Identity verification: No ### What 1inch is 1inch is a centralised exchange venue launched in 2020. At our last check it reported $31M of 24-hour turnover across 240 markets, with $1.6B in 30-day volume. ### Fees Spot trading is priced at 0% maker and 0% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. 1inch supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does 1inch charge per trade? A: 0% maker and 0% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does 1inch require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## SwissBorg URL: https://www.caijing188.com/exchange/swissborg/ Updated: 2026-10-27T20:02:26+08:00 Verified: 2026-10-27 ### Quick answer SwissBorg is a exchange venue with $21M in reported 24-hour volume and 96 markets as of October 2026. Spot trading costs 0.25% taker on the entry tier. Identity verification is yes. Headquartered in Switzerland. ### Key takeaways - $21M reported 24-hour volume - 96 listed markets - 0.25% taker fee, entry tier - Identity verification: Yes ### What SwissBorg is SwissBorg is a centralised exchange venue launched in 2017. At our last check it reported $21M of 24-hour turnover across 96 markets, with $1.1B in 30-day volume. ### Fees Spot trading is priced at 0.25% maker and 0.25% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Switzerland. Identity verification: Yes. Available in: Global, with restrictions in some markets. SwissBorg supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does SwissBorg charge per trade? A: 0.25% maker and 0.25% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does SwissBorg require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## BTCTurk URL: https://www.caijing188.com/exchange/btcturk/ Updated: 2026-10-27T20:02:25+08:00 Verified: 2026-10-27 ### Quick answer BTCTurk is a exchange venue with $50M in reported 24-hour volume and 383 markets as of October 2026. Spot trading costs 0.2% taker on the entry tier. Identity verification is yes. Headquartered in Türkiye. ### Key takeaways - $50M reported 24-hour volume - 383 listed markets - 0.2% taker fee, entry tier - Identity verification: Yes ### What BTCTurk is BTCTurk is a centralised exchange venue launched in Jul 2013. At our last check it reported $50M of 24-hour turnover across 383 markets, with $5.2B in 30-day volume. ### Fees Spot trading is priced at 0.2% maker and 0.2% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Türkiye. Identity verification: Yes. Available in: Global, with restrictions in some markets. BTCTurk supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does BTCTurk charge per trade? A: 0.2% maker and 0.2% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does BTCTurk require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Paribu URL: https://www.caijing188.com/exchange/paribu/ Updated: 2026-10-27T20:02:25+08:00 Verified: 2026-10-27 ### Quick answer Paribu is a exchange venue with $87M in reported 24-hour volume and 268 markets as of October 2026. Spot trading costs 0.2% taker on the entry tier. Identity verification is yes. Headquartered in Türkiye. ### Key takeaways - $87M reported 24-hour volume - 268 listed markets - 0.2% taker fee, entry tier - Identity verification: Yes ### What Paribu is Paribu is a centralised exchange venue launched in Dec 2016. At our last check it reported $87M of 24-hour turnover across 268 markets, with $5.1B in 30-day volume. ### Fees Spot trading is priced at 0.2% maker and 0.2% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Türkiye. Identity verification: Yes. Available in: Global, with restrictions in some markets. Paribu supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Paribu charge per trade? A: 0.2% maker and 0.2% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Paribu require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## CoinEx URL: https://www.caijing188.com/exchange/coinex/ Updated: 2026-10-27T20:02:24+08:00 Verified: 2026-10-27 ### Quick answer CoinEx is a exchange venue with $120M in reported 24-hour volume and 640 markets as of October 2026. Spot trading costs 0.2% taker on the entry tier. Identity verification is no. Headquartered in Hong Kong, China. ### Key takeaways - $120M reported 24-hour volume - 640 listed markets - 0.2% taker fee, entry tier - Identity verification: No ### What CoinEx is CoinEx is a centralised exchange venue launched in 2017. At our last check it reported $120M of 24-hour turnover across 640 markets, with $6.3B in 30-day volume. ### Fees Spot trading is priced at 0.2% maker and 0.2% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Hong Kong, China. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. CoinEx supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does CoinEx charge per trade? A: 0.2% maker and 0.2% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does CoinEx require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## Gemini URL: https://www.caijing188.com/exchange/gemini/ Updated: 2026-10-27T20:02:24+08:00 Verified: 2026-10-27 ### Quick answer Gemini is a exchange venue with $91M in reported 24-hour volume and 312 markets as of October 2026. Spot trading costs 0.4% taker on the entry tier. Identity verification is yes. Headquartered in United States. ### Key takeaways - $91M reported 24-hour volume - 312 listed markets - 0.4% taker fee, entry tier - Identity verification: Yes ### What Gemini is Gemini is a centralised exchange venue launched in Oct 2014. At our last check it reported $91M of 24-hour turnover across 312 markets, with $4.1B in 30-day volume. ### Fees Spot trading is priced at 0.2% maker and 0.4% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in United States. Identity verification: Yes. Available in: Global, with restrictions in some markets. Gemini supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Gemini charge per trade? A: 0.2% maker and 0.4% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Gemini require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Bitstamp URL: https://www.caijing188.com/exchange/bitstamp/ Updated: 2026-10-27T20:02:23+08:00 Verified: 2026-10-27 ### Quick answer Bitstamp is a exchange venue with $300M in reported 24-hour volume and 253 markets as of October 2026. Spot trading costs 0.5% taker on the entry tier. Identity verification is yes. Headquartered in Luxembourg. ### Key takeaways - $300M reported 24-hour volume - 253 listed markets - 0.5% taker fee, entry tier - Identity verification: Yes ### What Bitstamp is Bitstamp is a centralised exchange venue launched in Jul 2011. At our last check it reported $300M of 24-hour turnover across 253 markets, with $15.4B in 30-day volume. ### Fees Spot trading is priced at 0.3% maker and 0.5% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Luxembourg. Identity verification: Yes. Available in: Global, with restrictions in some markets. Bitstamp supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Bitstamp charge per trade? A: 0.3% maker and 0.5% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Bitstamp require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Uniswap URL: https://www.caijing188.com/exchange/uniswap/ Updated: 2026-10-27T20:02:23+08:00 Verified: 2026-10-27 ### Quick answer Uniswap is a exchange venue with $159M in reported 24-hour volume and 652 markets as of October 2026. Spot trading costs 0.3% taker on the entry tier. Identity verification is no. Headquartered in Decentralised. ### Key takeaways - $159M reported 24-hour volume - 652 listed markets - 0.3% taker fee, entry tier - Identity verification: No ### What Uniswap is Uniswap is a centralised exchange venue launched in 2018. At our last check it reported $159M of 24-hour turnover across 652 markets, with $8.2B in 30-day volume. ### Fees Spot trading is priced at 0.3% maker and 0.3% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Decentralised. Identity verification: No. Available in: Crypto-only, no identity check on the trading account. Uniswap supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Uniswap charge per trade? A: 0.3% maker and 0.3% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Uniswap require KYC? A: No for crypto-only trading. Fiat deposits generally require verification. ## Bithumb URL: https://www.caijing188.com/exchange/bithumb/ Updated: 2026-10-27T20:02:21+08:00 Verified: 2026-10-27 ### Quick answer Bithumb is a exchange venue with $370M in reported 24-hour volume and 489 markets as of October 2026. Spot trading costs 0.25% taker on the entry tier. Identity verification is yes. Headquartered in South Korea. ### Key takeaways - $370M reported 24-hour volume - 489 listed markets - 0.25% taker fee, entry tier - Identity verification: Yes ### What Bithumb is Bithumb is a centralised exchange venue launched in Jun 2016. At our last check it reported $370M of 24-hour turnover across 489 markets, with $7.4B in 30-day volume. ### Fees Spot trading is priced at 0.25% maker and 0.25% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in South Korea. Identity verification: Yes. Available in: Global, with restrictions in some markets. Bithumb supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Bithumb charge per trade? A: 0.25% maker and 0.25% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Bithumb require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Crypto.com URL: https://www.caijing188.com/exchange/crypto-com/ Updated: 2026-10-27T20:02:20+08:00 Verified: 2026-10-27 ### Quick answer Crypto.com is a exchange venue with $640M in reported 24-hour volume and 562 markets as of October 2026. Spot trading costs 0.075% taker on the entry tier. Identity verification is yes. Headquartered in Singapore. ### Key takeaways - $640M reported 24-hour volume - 562 listed markets - 0.075% taker fee, entry tier - Identity verification: Yes ### What Crypto.com is Crypto.com is a centralised exchange venue launched in 2019. At our last check it reported $640M of 24-hour turnover across 562 markets, with $29.1B in 30-day volume. ### Fees Spot trading is priced at 0.075% maker and 0.075% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Singapore. Identity verification: Yes. Available in: Global, with restrictions in some markets. Crypto.com supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Crypto.com charge per trade? A: 0.075% maker and 0.075% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Crypto.com require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Upbit URL: https://www.caijing188.com/exchange/upbit/ Updated: 2026-10-27T20:02:19+08:00 Verified: 2026-10-27 ### Quick answer Upbit is a exchange venue with $920M in reported 24-hour volume and 875 markets as of October 2026. Spot trading costs 0.25% taker on the entry tier. Identity verification is yes. Headquartered in South Korea. ### Key takeaways - $920M reported 24-hour volume - 875 listed markets - 0.25% taker fee, entry tier - Identity verification: Yes ### What Upbit is Upbit is a centralised exchange venue launched in Oct 2017. At our last check it reported $920M of 24-hour turnover across 875 markets, with $51.4B in 30-day volume. ### Fees Spot trading is priced at 0.25% maker and 0.25% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in South Korea. Identity verification: Yes. Available in: Global, with restrictions in some markets. Upbit supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Upbit charge per trade? A: 0.25% maker and 0.25% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Upbit require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Bitget URL: https://www.caijing188.com/exchange/bitget/ Updated: 2026-10-27T20:02:19+08:00 Verified: 2026-10-27 ### Quick answer Bitget is a exchange venue with $940M in reported 24-hour volume and 1,120 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is yes. Headquartered in Seychelles. ### Key takeaways - $940M reported 24-hour volume - 1,120 listed markets - 0.1% taker fee, entry tier - Identity verification: Yes ### What Bitget is Bitget is a centralised exchange venue launched in 2018. At our last check it reported $940M of 24-hour turnover across 1,120 markets, with $74.6B in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Seychelles. Identity verification: Yes. Available in: Global, with restrictions in some markets. Bitget supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Bitget charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Bitget require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## KuCoin URL: https://www.caijing188.com/exchange/kucoin/ Updated: 2026-10-27T20:02:18+08:00 Verified: 2026-10-27 ### Quick answer KuCoin is a exchange venue with $1.02B in reported 24-hour volume and 970 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is yes. Headquartered in Seychelles. ### Key takeaways - $1.02B reported 24-hour volume - 970 listed markets - 0.1% taker fee, entry tier - Identity verification: Yes ### What KuCoin is KuCoin is a centralised exchange venue launched in Aug 2017. At our last check it reported $1.02B of 24-hour turnover across 970 markets, with $58.9B in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Seychelles. Identity verification: Yes. Available in: Global, with restrictions in some markets. KuCoin supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does KuCoin charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does KuCoin require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Kraken URL: https://www.caijing188.com/exchange/kraken/ Updated: 2026-10-27T20:02:18+08:00 Verified: 2026-10-27 ### Quick answer Kraken is a exchange venue with $980M in reported 24-hour volume and 1,407 markets as of October 2026. Spot trading costs 0.26% taker on the entry tier. Identity verification is yes. Headquartered in United States · global. ### Key takeaways - $980M reported 24-hour volume - 1,407 listed markets - 0.26% taker fee, entry tier - Identity verification: Yes ### What Kraken is Kraken is a centralised exchange venue launched in Jul 2011. At our last check it reported $980M of 24-hour turnover across 1,407 markets, with $55.3B in 30-day volume. ### Fees Spot trading is priced at 0.16% maker and 0.26% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in United States · global. Identity verification: Yes. Available in: Global, with restrictions in some markets. Kraken supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Kraken charge per trade? A: 0.16% maker and 0.26% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Kraken require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Coinbase URL: https://www.caijing188.com/exchange/coinbase/ Updated: 2026-10-27T20:02:17+08:00 Verified: 2026-10-27 ### Quick answer Coinbase is a exchange venue with $2.05B in reported 24-hour volume and 917 markets as of October 2026. Spot trading costs 0.6% taker on the entry tier. Identity verification is yes. Headquartered in United States. ### Key takeaways - $2.05B reported 24-hour volume - 917 listed markets - 0.6% taker fee, entry tier - Identity verification: Yes ### What Coinbase is Coinbase is a centralised exchange venue launched in 2012. At our last check it reported $2.05B of 24-hour turnover across 917 markets, with $138B in 30-day volume. ### Fees Spot trading is priced at 0.4% maker and 0.6% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in United States. Identity verification: Yes. Available in: Global, with restrictions in some markets. Coinbase supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Coinbase charge per trade? A: 0.4% maker and 0.6% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Coinbase require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Gate.io URL: https://www.caijing188.com/exchange/gate-io/ Updated: 2026-10-27T20:02:17+08:00 Verified: 2026-10-27 ### Quick answer Gate.io is a exchange venue with $1.31B in reported 24-hour volume and 1,748 markets as of October 2026. Spot trading costs 0.15% taker on the entry tier. Identity verification is yes. Headquartered in Cayman Islands. ### Key takeaways - $1.31B reported 24-hour volume - 1,748 listed markets - 0.15% taker fee, entry tier - Identity verification: Yes ### What Gate.io is Gate.io is a centralised exchange venue launched in Jan 2013. At our last check it reported $1.31B of 24-hour turnover across 1,748 markets, with $66.2B in 30-day volume. ### Fees Spot trading is priced at 0.15% maker and 0.15% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Cayman Islands. Identity verification: Yes. Available in: Global, with restrictions in some markets. Gate.io supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Gate.io charge per trade? A: 0.15% maker and 0.15% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Gate.io require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Bybit URL: https://www.caijing188.com/exchange/bybit/ Updated: 2026-10-27T20:02:16+08:00 Verified: 2026-10-27 ### Quick answer Bybit is a exchange venue with $2.74B in reported 24-hour volume and 527 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is yes. Headquartered in Dubai · global. ### Key takeaways - $2.74B reported 24-hour volume - 527 listed markets - 0.1% taker fee, entry tier - Identity verification: Yes ### What Bybit is Bybit is a centralised exchange venue launched in 2018. At our last check it reported $2.74B of 24-hour turnover across 527 markets, with $156B in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Dubai · global. Identity verification: Yes. Available in: Global, with restrictions in some markets. Bybit supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Bybit charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Bybit require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## OKX URL: https://www.caijing188.com/exchange/okx/ Updated: 2026-10-27T20:02:16+08:00 Verified: 2026-10-27 ### Quick answer OKX is a exchange venue with $3.18B in reported 24-hour volume and 1,210 markets as of October 2026. Spot trading costs 0.08% taker on the entry tier. Identity verification is yes. Headquartered in Seychelles · global. ### Key takeaways - $3.18B reported 24-hour volume - 1,210 listed markets - 0.08% taker fee, entry tier - Identity verification: Yes ### What OKX is OKX is a centralised exchange venue launched in Jan 2014. At our last check it reported $3.18B of 24-hour turnover across 1,210 markets, with $198B in 30-day volume. ### Fees Spot trading is priced at 0.08% maker and 0.08% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Seychelles · global. Identity verification: Yes. Available in: Global, with restrictions in some markets. OKX supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does OKX charge per trade? A: 0.08% maker and 0.08% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does OKX require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Binance URL: https://www.caijing188.com/exchange/binance/ Updated: 2026-10-27T20:02:15+08:00 Verified: 2026-10-27 ### Quick answer Binance is a exchange venue with $6.42B in reported 24-hour volume and 1,379 markets as of October 2026. Spot trading costs 0.1% taker on the entry tier. Identity verification is yes. Headquartered in Global (restricted in some markets). ### Key takeaways - $6.42B reported 24-hour volume - 1,379 listed markets - 0.1% taker fee, entry tier - Identity verification: Yes ### What Binance is Binance is a centralised exchange venue launched in Jul 2017. At our last check it reported $6.42B of 24-hour turnover across 1,379 markets, with $394B in 30-day volume. ### Fees Spot trading is priced at 0.1% maker and 0.1% taker on the published entry tier. Deposit fees are usually zero; the withdrawal fee depends on the network you use and is the number that dominates small transfers. ### Access and custody Headquartered in Global (restricted in some markets). Identity verification: Yes. Available in: Global, with restrictions in some markets. Binance supports fiat on-ramps in major currencies where it is licensed. ### What to check before you deposit Confirm the legal entity on your statement, test a withdrawal before you size a position, and keep a record of the cost basis at the moment any reward is credited. *Figures re-checked 2026-10-27. Volume is reported by the venue and can be inflated; we do not adjust it.* ### FAQ Q: What does Binance charge per trade? A: 0.1% maker and 0.1% taker on the published entry tier. VIP tiers reduce this with volume. Q: Does Binance require KYC? A: Yes. Identity verification is required before trading or withdrawing. ## Impermanent loss, worked through with one pool and one price move URL: https://www.caijing188.com/impermanent-loss-one-pool/ Updated: 2026-10-27T22:09:40+08:00 Verified: 2026-10-27 ### Quick answer Impermanent loss is the gap between a pool position and simply holding, not a loss against your deposit. In a 50/50 pool where one asset doubles, the position ends 5.72% below holding, as of October 2026. Fees must beat that gap, and the quoted fee APR is trailing rather than forward. ### Key takeaways - A double in one asset costs a 50/50 pool 5.72% against holding, on a worked example dated October 2026 - Divergence reaches 13.40% at a triple and is symmetric, so a 50% drop costs the same 5.72% - Covering 5.72% at a 10% fee APR takes about 6.9 months of fees - Fee APR on a pool page is trailing and is usually divided by a TVL that already fell ![A two-asset pool rebalancing after a single price move](https://img1.ladyww.cn/picture/Picture00431.jpg) Impermanent loss is not a loss in dollars. It is the gap between two outcomes measured at the same moment: the position inside the pool, and the position you would still hold had you never deposited. One pool and one price move is enough to see it. ## Set the pool before you set the price A constant-product pool holds two assets and keeps their product constant. Deposit 1 ETH and 2,000 USDC when ETH trades at 2,000, and the pool holds 1 ETH against 2,000 USDC, so k is 2,000. The split is even because the curve prices both sides from one formula, and that is the only starting point with zero divergence. ## The price doubles and the pool sells into it ETH goes to 4,000. The pool cannot hold its ETH, because anyone may buy at the curve price, and the only way to keep k constant is to give up ETH and take in USDC. Solving the curve at the new price leaves 0.7071 ETH and 2,828.43 USDC, worth 5,656.85 USDC. Holding would have left 1 ETH and 2,000 USDC, worth 6,000. The gap is 343.15 USDC, or 5.72% of the holding value. Along the way the pool sold 0.2929 ETH for 828.43 USDC, an average of about 2,828 against a final price of 4,000. It is a mechanical seller into strength, paid the average rather than the top. ## The loss is relative to holding The position did not lose money. It started at 4,000 and finished at 5,656.85, up 41.4%, while holding finished up 50%. That is the whole definition: divergence compares the pool with the hold at one timestamp, which is why a pool can be profitable in dollars and still underperform the asset in it. It also explains the word impermanent. If the price returns to 2,000, the pool returns to 1 ETH and 2,000 USDC and the gap closes to zero, less whatever fees accrued. The number only becomes permanent when you withdraw at the wrong moment. ## The formula, and why doubling is the reference case For a 50/50 pool and a price ratio r, divergence is 2 times the square root of r, divided by 1 plus r, minus 1. The function is symmetric in the ratio, so a double and a halving cost the same. | Price ratio r | Pool value vs holding | Divergence | |—|—|—| | 1.25 (+25%) | 0.9938 | -0.62% | | 1.50 (+50%) | 0.9798 | -2.02% | | 2.00 (+100%) | 0.9428 | -5.72% | | 3.00 (+200%) | 0.8660 | -13.40% | | 0.50 (-50%) | 0.9428 | -5.72% | Derived from the constant-product formula applied to the worked example, October 2026 pricing. The shape matters more than any row. Divergence grows slowly and then accelerates, so a pool pairing one volatile asset with a stablecoin is a position against large moves in either direction. ## Fees have to beat the divergence Fees are the only offset. They accrue into pool value rather than into a separate balance, so they appear as a pool value slightly better than the curve predicts. Break-even is arithmetical. Cumulative fees must exceed the divergence, as a percentage of position value, over the same window. At fee APR f, the months needed to cover 5.72% are roughly 5.72 divided by f, times 12: about 6.9 months at 10%, about 3.4 at 20%. No pool quotes a forward fee APR. The number on the page is realised fees over the last 7 or 30 days, annualised, and denominated in whichever asset the pool holds at the snapshot. Volume follows volatility, and volatility is the condition that produces divergence, so the trailing APR and the divergence tend to peak together. The figure used to justify a deposit is measured in the period least like the one you will hold through. It is also divided by pool TVL, which includes the value that just fell, so the same fees look better after a drawdown. TVL is a deposit count, not a revenue line ([fee revenue is not TVL](/l2-fee-revenue-not-tvl/)), and we treat published fee pages as trailing evidence for the same reason ([fee pages change monthly](/fee-pages-change-monthly/)). ## Verdict Run three numbers before depositing: divergence at the move you consider plausible, fee income you can document over a comparable window, and the fee APR restated on your entry size rather than today’s TVL. If the third does not clearly beat the first, the pool is a way to hold two assets with a tax on conviction. If you would not hold both assets through that move, the fee is not compensation. It is a subsidy for the other side of your exit. ### FAQ Q: Is impermanent loss a real loss if the position is still up in dollars? A: It is real in the only sense that matters for comparison: you finish with fewer dollars than the same assets held outside the pool. In the worked example the position rose 41.4% while holding rose 50%. The gap is 343.15 USDC, and it becomes permanent when you withdraw at that price. Q: Why does a 50% drop cost the same as a double? A: Because divergence depends on the ratio between the two assets, and a ratio of 2 and a ratio of 0.5 are mirror images on the same curve. Both leave the pool holding 0.7071 of the volatile asset, so both produce a 5.72% gap against holding. Q: Do fees always cover impermanent loss? A: No, and no pool page can tell you in advance. Fees accrue with volume, volume follows volatility, and volatility is the condition that produces divergence. The two tend to peak together, so a high trailing fee APR is often measured during the period that generated the loss. ## What a listing announcement actually moves, and how fast URL: https://www.caijing188.com/listing-announcement-liquidity/ Updated: 2026-10-27T22:09:40+08:00 Verified: 2026-10-27 ### Quick answer A listing announcement moves price in minutes and book depth over days, so day-one volume mostly reflects market makers and bots, as of October 2026. Our board ranks 24 venues, and FeeEdge's 17 August 2026 survey put entry-tier spot taker fees near 0.10%. Check depth at your size before you size. ### Key takeaways - Announcement-day volume measures inventory that crossed, not liquidity that stayed - FeeEdge's 17 August 2026 survey put entry-tier spot taker fees near 0.10% - L1 ERC-20 withdrawals cost $2 to $10 on the same survey, above $50 in congestion - The board ranks 24 venues on sampled 24h volume as of October 2026, with the top row near a third of the sample ![A candlestick spike beside an order book depth ladder](https://img1.ladyww.cn/picture/Picture00401.jpg) A listing announcement moves two things at once: price and depth. The price move is visible within minutes and gets screenshotted. The depth change takes days and rarely gets recorded. For anyone who intends to trade the new pair rather than watch it, depth is the number that decides whether the visible move was ever capturable. ## The announcement trade is mostly not yours When a venue lists a pair, it has to seed a book before organic flow exists. Market makers quote first, usually under an agreement that obliges them to hold a two-sided market for some period. Arbitrage bots arrive next, because the asset already trades elsewhere and a new listing is a price gap with a ticker attached. Both groups generate volume. Neither is obliged to stay. A maker’s obligation usually specifies a minimum size and a maximum spread, and that minimum is typically far below what a large order needs. A bot has no obligation beyond its own profitability. So the day-one volume print measures how much inventory crossed the book, not how much liquidity remained in it. Those are different numbers, and only one of them appears on a ranking. ## The spread narrows over days A new pair usually opens wider than the venue’s mature pairs. The quoting desk is carrying inventory risk in an asset whose local price history is measured in hours, so it prices that risk into the quote. The spread then compresses as competing makers arrive. That compression runs on a clock measured in days. It is also not smooth. The first step comes when a second maker joins. The second comes when the venue’s fee tier attracts flow that makers can warehouse. Neither event happens on announcement day. Fees are a separate axis and a listing does not move them. FeeEdge’s survey of 17 August 2026 put entry-tier spot taker fees near 0.10% at most venues, and put exchange withdrawal fees at roughly 0.5 to 1 USDT for Solana SPL, about 0.3 USDT for BEP-20, and highest for ERC-20. The same survey put L1 ERC-20 transfers at $2 to $10, and above $50 in congestion. Those are the costs of a round trip before any slippage. A listing announcement changes none of them. ## Depth is the part the announcement does not fix Depth inside one percent is the size you can trade before the mid moves a percent. It is what a slippage model needs and what a volume column cannot supply. A venue can print a large volume and stay thin. Volume counts turnover in both directions, includes makers trading with each other, and resets every 24 hours. Depth is a snapshot of resting orders at this moment, which describes willingness rather than activity. Our own board shows this in its structure rather than in any single row. It ranks 24 venues on sampled 24h volume as of October 2026, and the top row carries roughly a third of that sample. When one venue absorbs a third, the rest of the table sits in a long tail whose books are much thinner than the ranking implies. ## What to check before size Cheap checks come before the order. Expensive ones come after it. The sequence is the whole point. Start with the book at your size, not at the touch. Take the notional you intend to trade and walk the ladder until it fills. The average fill price on that walk is your slippage estimate, and it will usually differ from the touch by more than a model that reads volume suggests. Then check who is quoting. Two makers at the touch is a book that can vanish. Six is a book that has to be replaced by someone else before it thins. Then check the exit. A position is only as liquid as the path out of it, and the exit cost is published. The board’s fee pages are a reasonable place to start, because venue fees change more often than traders assume. | Signal | What it measures | When it holds | |—|—|—| | Quoted spread | Cost of a small crossing order | Only at the size the maker is obliged to show | | Depth inside 1% | Size tradeable before a one percent move | At snapshot time, not at settlement | | 24h volume | Turnover across the window | Never as a proxy for resting size | | Maker count at the touch | How many independent quotes stand behind the book | As a lower bound, since one desk can run several | | Withdrawal fee | Cost of leaving the venue | Until the next fee page change | ## Verdict The tradable part of a listing is depth, and depth arrives on a slower schedule than the price. Treat announcement-day volume as evidence that inventory moved, not that a market exists. Check the book at your size before you size. Read the [monthly fee page changes](/fee-pages-change-monthly/) for the exit cost, and use the [venue directory](/exchange/) to compare how many books an order would be split across. A listing is a new venue for a coin, not a new pool of liquidity for it. ### FAQ Q: Why does volume spike on listing day if liquidity does not? A: Volume counts turnover, and turnover can come from participants who have no intention of holding inventory. Market makers seed a new book under an obligation with a minimum size, and arbitrage bots trade the new pair against the same asset elsewhere. Both print volume on day one and both can be gone by day three. Q: How long does a new pair take to trade like a mature one? A: In practice the spread compresses over days rather than hours, as a second and third maker arrive and as the venue's fee tier attracts flow that makers can warehouse. There is no fixed schedule, and some pairs never reach the depth of the venue's established books. Q: What is the cheapest check before buying a newly listed pair? A: Walk the ladder at the notional you actually intend to trade and compare the average fill with the touch price. That single number captures spread and slippage together, and it costs nothing but a few seconds before the order. ## MiCA passporting: which entity actually holds your account URL: https://www.caijing188.com/mica-entity-passporting/ Updated: 2026-10-27T22:09:41+08:00 Verified: 2026-10-27 ### Quick answer A passporting licence lets one authorised firm serve several markets, so the contracting entity in your terms can differ from the brand on the homepage, as of October 2026. Our board ranks 24 venues, and reserve disclosures remain snapshots rather than guarantees. Read the client agreement, the register entry and the reserve page, in that order. ### Key takeaways - One licence can serve several markets, so the entity on your statement can differ from the brand - The counterparty is named in the client agreement, the register entry and the reserve page, and the three should match - A licence is not insurance, and deposit protection schemes generally cover bank deposits, not crypto balances - Our board ranks 24 venues as of October 2026, with the top row near a third of the sampled 24-hour volume ![A passport on a desk beside printed account statements](https://img1.ladyww.cn/picture/Picture00421.jpg) One licence can serve several markets. That is what passporting is for, and it means the brand you clicked is not automatically the company that holds your balance. The counterparty on your statement is a legal entity with its own registration number, its own supervisor and its own place in an insolvency queue. For a reader with a position open, that entity is the risk, not the logo. ## Passporting is a notification, not a second licence A passporting arrangement lets a firm authorised in one jurisdiction provide the same service in another after a notification, rather than a fresh authorisation. The home supervisor keeps supervising. The host supervisor is told, and can act, but the licence does not arrive as a new document. The commercial logic is straightforward. One compliance function, one treasury, several markets. The customer-facing logic is the part that gets lost, because a single interface then has to represent several possible contracting parties. Which entity you face usually depends on where the account was opened and which service you are using. Two readers on the same homepage can be customers of two different companies with two different regulators. ## The entity is named in three documents Three documents settle the question, and all three are usually public. None of them is the homepage. ### The client agreement The terms you accepted name the counterparty, usually in the first two pages and again in a definitions schedule. Look for the registered name, the company number and the registered office. A trading name in the header is marketing. The registered name is the party that owes you the balance. If the agreement incorporates separate product terms, check which entity appears in those too. A group can hold the account in one company and book the earn product in another. ### The public register Most licensing regimes publish a register of authorised firms. The entry should match the registered name in your terms, not the brand. Where the register lists permitted activities, check that custody and fiat transfer appear. A firm can be authorised for one service and not another. The interface will not tell you which one you are using, and the answer can differ between the spot balance and a savings product. ### The reserve disclosure Reserve reporting is a snapshot, not a guarantee. We treat a disclosure as evidence of a balance at a stated time, and [the three numbers that matter](/proof-of-reserves-three-numbers/) sets out what such a report can support. The question to add here is whose balance sheet the report covers. If the entity holding your account is not the entity in the report, the report describes someone else’s assets. ## Why the entity decides the outcome Complaint routes follow the licence. A dispute goes to the supervisor named on the register, not to the regulator of the country you happen to live in. That can mean a process in a language you do not speak, on a timetable you did not choose. Segregation and custody rules are set by the licensing jurisdiction. What happens to client assets in a wind-down is a question about that jurisdiction’s insolvency law, and about whether the assets were segregated before the wind-down started. Deposit protection, where it exists, generally covers bank deposits rather than crypto balances. A licence is not insurance. ## A reading order that takes ten minutes Open the client agreement, search for the word registered, and write the legal name down. Then open the register and confirm the name appears with the services you use. Then open the reserve page and check whose balance sheet it describes. Our board ranks 24 venues as of October 2026, and the top row carries roughly a third of the sampled 24-hour volume. Size is not clarity. A large venue can still route you to a small entity, and the small entity is the one in the queue. | Document | What to look for | Why it decides your position | |—|—|—| | Client agreement | Registered name, company number, registered office | Names the counterparty that owes you the balance | | Public register | The same name, plus the activities you use | Sets the supervisor and the complaint route | | Reserve disclosure | Whose balance sheet the report covers | A snapshot for one entity, not for the group | | Fee schedule | Which entity bills the fee | Shows which entity your order was booked with | ## What to do next Write the legal name in your notes next to the account, the way you would note a bank’s sort code. When the interface changes, the entity sometimes changes with it, and the change arrives in the terms rather than on a landing page. Keep the version you accepted. A dated copy is the difference between an argument about what the contract said and a check of what it says now. For the wider comparison, our [venue board](/exchange/) lists 24 exchanges as of October 2026 and the [methodology note](/methodology/) explains what the columns measure. Neither substitutes for the register entry, which is the only document that names a supervisor. The check is cheap because the information is public. The failure mode is expensive, because you discover the counterparty at the moment you need to enforce something against it. ### FAQ Q: If a venue is licensed in the EU, is my account held by that licensed company? A: Not necessarily. Under a passporting framework a group can serve several markets from one authorisation, but the account itself sits with whichever entity signed your terms. That entity may be the licensed one, a branch of it, or an affiliate in another jurisdiction. The name appears in the client agreement and should also appear on the public register. Q: Does a register entry mean my balance is safe? A: No. A register entry records that a firm is authorised for certain activities and names the supervisor. It says nothing about solvency, and it is not a guarantee. Segregation rules and insolvency treatment are set by the licensing jurisdiction, which is a separate question from whether the licence exists. Q: What is the fastest way to check which entity I am dealing with? A: Search the client agreement for the word registered and note the full legal name, company number and registered office. Then confirm that name, not the brand, appears on the public register with the activities you actually use. Finally check whose balance sheet the reserve page describes. ## Blob space after EIP-4844: what an L2 does when data gets expensive URL: https://www.caijing188.com/blob-space-pricing/ Updated: 2026-10-27T22:09:42+08:00 Verified: 2026-10-27 ### Quick answer Blob space decoupled L2 data cost from mainnet gas, but it is its own market with its own price. Ethereum.org's snapshot, as of October 2026, listed Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction. When the lane is contested a rollup either pays more or posts less often, and both reach the user. ### Key takeaways - Ethereum.org's 1 to 6 October 2026 snapshot put Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction - The same snapshot showed OP Mainnet at $0.00, Scroll at $0.002 to $0.003 and Linea near $0.024 - A rollup facing a higher blob fee can pay more or post less often, and both push cost toward users - Exit clocks run 7 days on Base and OP Mainnet and about 6.4 days on Arbitrum, per Ethereum.org in October 2026 ![Blob fee curve beside two rollup fee rows](https://img1.ladyww.cn/picture/Picture00411.jpg) EIP-4844 gave rollups a data lane of their own, and with it a separate market for space. Before the change, a rollup’s batch data competed with ordinary mainnet traffic for the same gas. After it, the batch pays a blob fee that moves on its own curve. That decoupling is why a rollup can quote a fee in fractions of a cent while mainnet gas does something else, and why that fee can rise while mainnet is quiet. ## What blobs actually changed EIP-4844 introduced a data type that rollups post to Ethereum and that the protocol prices separately from execution gas. The headline effect is a discount. The structural effect is that data cost became a market with its own supply and its own fee, and that fee responds to how full recent blocks were. For a rollup, that turns data cost into a variable it can influence but does not set. When the lane is quiet, posting a batch is cheap and the user fee can sit near zero. When several rollups want the lane at once, the price of posting rises and the operator has to decide what to do about it. ## Two levers, and both reach the user An L2 facing a higher blob fee has two options. The first is to pay. The rollup absorbs a higher data cost and either narrows its margin or raises the fee it charges. Most rollups set that fee by policy rather than by auction, so pass-through is delayed and lumpy. Users see a change days after the blob market moved. The second is to post less often. A batch can cover more transactions if the operator waits longer before posting, which spreads one expensive posting across more user activity. The cost of that choice is time. The data behind your transaction reaches mainnet later, and the window in which your L1 claim depends on the operator’s word gets longer. Neither lever is visible in a fee number by itself. One shows up as a higher fee, the other as a cheap fee on a chain that settles more slowly. ## The fee page is the output, not the input Which lever a rollup pulled is partly readable in the fee it publishes. Ethereum.org’s L2 directory, in its 1 to 6 October 2026 snapshot, listed Base at $0.001 to $0.002 per transaction and Arbitrum One at $0.004 to $0.005, while OP Mainnet displayed $0.00, Scroll showed $0.002 to $0.003, Starknet $0.005 to $0.008 with gas paid in STRK, and Linea about $0.024. A spread that wide is not explained by blob fees alone: fee policy, proof system, batch cadence and payment asset all differ. What the comparison does establish is that the user-facing price is a decision layered on top of a data cost, not a readout of it. Scale is the other half. The same directory, in its October 2026 reading, put Base at $14.6B in value and Arbitrum One at $12.3B, with OP Mainnet at $1.65B, Starknet at $389M, Linea at $352M, ZKsync Era at $227M and Scroll at $45.7M. A larger chain spreads a fixed posting cost across more transactions, which is one reason the biggest names quote the smallest per-transaction numbers. | Chain | Fee per transaction, Ethereum.org, 1 to 6 October 2026 | What the figure does not show | |—|—|—| | Base | $0.001 to $0.002 | Batch cadence behind the fee | | Arbitrum One | $0.004 to $0.005 | Length of the dispute window | | OP Mainnet | $0.00 displayed | Whether the fee is subsidised | | Scroll | $0.002 to $0.003 | Thin flow under a small base | | Starknet | $0.005 to $0.008, paid in STRK | Token price exposure | | Linea | About $0.024 | Why it sits at the top of the range | ## What the fee number does not carry Three things sit behind a published figure. Batch frequency. A cheap fee can come from posting rarely rather than from cheap data. Cadence is the number that tells you which, and it is usually not on the fee page. The exit. A fee is the cost of entering an economy, not of leaving it. Base and OP Mainnet withdrawals to mainnet take 7 days, Arbitrum’s dispute window is 45,818 blocks, about 6.4 days, and ZKsync Era quotes a 3-hour minimum, per Ethereum.org’s L2 directory in October 2026. On a cheap chain, that is where the real cost usually sits. The payment asset. Starknet’s fee is denominated in STRK in that snapshot, so the dollar figure depends on a token price as well as on gas. ## What to do next Read the fee as a lagging indicator. It is set after the data market moves, so the number on screen describes the last decision, not the next one. Then check the two things the fee page leaves out: how often the chain posts, and how long the exit takes. The [fee to revenue question](/l2-fee-revenue-not-tvl/) decides whether a low fee is a subsidy or an efficiency, and [what a sequencer does](/what-a-sequencer-does/) explains who sets the price you are quoted. Compare chains by exit, not by entry. Entering is cheap almost everywhere. Leaving is where the models differ. ### FAQ Q: Does a cheap fee mean a rollup is efficient or subsidised? A: The fee page cannot tell you. Ethereum.org's 1 to 6 October 2026 snapshot showed Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005, but the published number is set by the operator's policy on top of its data cost. A fee below cost is a subsidy, and it lasts as long as the operator wants it to. Q: Why would a rollup post data less often instead of paying more? A: Posting less often spreads one expensive batch across more user transactions, which lowers the per-transaction cost immediately. The price is latency: the data behind your transaction reaches mainnet later, and the window where your L1 claim depends on the operator's word grows longer. It is a cost transfer, not a saving. Q: Is my transaction final once the rollup shows it? A: No. The rollup's own ledger is a promise, and the mainnet claim arrives after its data and proof settle. Base and OP Mainnet quote 7 days for a withdrawal to mainnet, Arbitrum's dispute window is 45,818 blocks, about 6.4 days, and ZKsync Era quotes a 3-hour minimum, per Ethereum.org's L2 directory in October 2026. ## Hardware wallets: what the secure element does not protect URL: https://www.caijing188.com/secure-element-limits/ Updated: 2026-10-27T22:09:42+08:00 Verified: 2026-10-27 ### Quick answer The secure element keeps the key inside the chip. It does not read the transaction for you. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026, moved by signatures. If the device shows a hash instead of an amount and a destination, the chip is protecting a key you are spending blind. ### Key takeaways - The chip stops key extraction, within the PIN attempt limit, and nothing else - A device that signs faithfully signs a hostile payload as faithfully as a friendly one - Firmware is part of custody: check signed images and the on-device version, not the app - An ERC-20 withdrawal on Ethereum L1 cost $2 to $10 in FeeEdge's 17 August 2026 sample, which is the price of moving a position to self-custody ![Hardware wallet screen showing a decoded transfer](https://img1.ladyww.cn/picture/Picture00451.jpg) A secure element is the part of a hardware wallet that holds key material and refuses to hand it over. It is a real defence, and it is narrowly scoped: it protects the key from extraction, not the balance from a transaction you approved. The screen and the parsed amount are the boundary that actually decides whether funds leave. Firmware delivery and blind signing are the surfaces the chip never covers. ## What the chip guarantees The secure element generates the key inside its own boundary and signs inside it. The private key is never exposed to the host computer, so malware on a laptop cannot copy it, and a stolen device yields nothing without the PIN. Attempt counters, tamper detection and side-channel hardening belong to the same layer. That guarantee is about confidentiality of the key. It says nothing about authorisation. A device that signs faithfully will sign a hostile payload just as faithfully as a friendly one, because the chip has no opinion about intent. It only knows that a PIN was entered and a button was pressed. ## The signature is the transaction Every transaction is a message plus a signature. The chip’s job is to produce that signature correctly. The security question moves one step earlier: who decided the message was right, and how did you check it. This is the same on every account chain. A transfer, an approval, a permit, a contract call: all of them are payloads, and the device either renders the payload or it does not. A secure element behind an unreadable screen is an expensive way to sign things you cannot see. ## The screen is the real boundary ### What a screen has to show Four fields carry the decision: the amount, the destination in full, the fee, and the function called when the payload is not a plain transfer. For an approval, the spender matters more than the amount, because the spender receives the permission. ### Where truncation hides the attack Shortened addresses are the standard trick. A device showing the first four and last four characters is showing a label, not an address, and a lookalike can match both ends. Checking the middle closes it, which needs a screen that can display the middle. The pattern is covered in [the address poisoning habit](/address-poisoning-habit/). Decimals are the second trick. A token that reports 18 decimals when you expect 6 turns a modest transfer into something else on screen. A device that renders raw units has moved the arithmetic to the least reliable part of the system, which is you at one in the morning. ## Firmware is a supply chain A hardware wallet runs code, and that code arrives by update. The update path is as much a part of custody as the chip. Three checks matter, and none require a brand name. Does the vendor sign firmware and publish the verification? Can you confirm the version on the device itself rather than in the companion app? Does the device refuse to sign when the firmware image fails its check? A device bought second hand is a separate question. Unless you wipe it, generate a fresh seed on it and verify its firmware, you trust whoever held it last. Treat a used device as compromised. | Threat | Does the secure element stop it | Where the control actually sits | |—|—|—| | Key extraction from a stolen device | Yes, inside the PIN attempt limit | Device hardware and PIN policy | | Malware reading the seed on a laptop | Yes, if the seed was generated on device | Setup routine: never type a seed into a host | | Approval of a malicious transaction | No | Screen, decoded amount, destination | | Lookalike destination address | No | Full address display, middle characters | | Malicious firmware update | Partly, if the image is signed and checked on device | Vendor signing and the update path | | Hash-only blind signing | No | Refusing payloads the device cannot decode | ## What remains after the chip Blind signing. When the device cannot decode a payload it shows a digest, and a digest cannot be checked against intent. The fallback exists for complex payloads, and it is also the quiet path around every control above. Then coercion and loss. A chip cannot tell a signature given under duress from one given freely, and it cannot restore a seed you never wrote down. ## What to do next Buy the smallest number of devices that matches the size you hold. One for savings that never talks to a browser session full of unknown contracts, one cheaper setup for daily spending, and no shared seed. The [hot wallet sizing question](/how-much-in-a-hot-wallet/) is the same trade at a smaller scale. Then budget the move. FeeEdge’s 17 August 2026 sample put an ERC-20 withdrawal on Ethereum L1 at $2 to $10, while Solana SPL withdrawals cost 0.5 to 1 USDT. Paying that once to move a position is cheap. Paying it in a hurry is not a plan. Finally, decide the policy before the device arrives. What you will never sign, what you will always read on the screen, and which device holds the key allowed to approve contracts. The [reserve reporting question](/proof-of-reserves-three-numbers/) covers the exchange side of the same decision. ### FAQ Q: Does a secure element stop a drainer? A: No. A drainer needs a signature, not a key, and the secure element produces signatures on request. It will refuse to export the key, and it will sign an unlimited approval to a hostile contract if that approval is what you confirm on the device. The screen and the decoded amount are what stop it. Q: Is a second-hand hardware wallet safe if I reset it? A: Only after you wipe it, generate a fresh seed on the device and verify its firmware. A reset device with a seed you generated is functionally new. A device with someone else's seed, or with firmware you did not check, is not yours to trust. Q: Why does the screen matter more than the certification? A: Certification says the key is hard to extract. The screen says what you are authorising. Every transaction is a message plus a signature, so an unreadable message moves the decision to the companion app, which runs on a general purpose computer the device exists to distrust. ## Revenue per transaction is the L2 metric that survives incentives URL: https://www.caijing188.com/revenue-per-transaction/ Updated: 2026-10-27T22:09:43+08:00 Verified: 2026-10-27 ### Quick answer Base charged $0.001 to $0.002 per transfer and Arbitrum One $0.004 to $0.005 in Ethereum.org's 1 to 6 October 2026 snapshot, as of October 2026. Revenue per transaction turns a headline into a transfer count: a million dollars is 500 million to 1 billion transfers on Base and 200 to 250 million on Arbitrum. ### Key takeaways - Base charged $0.001 to $0.002 per transfer, Arbitrum One $0.004 to $0.005, on 1 to 6 October 2026 - $1M of revenue implies 500 million to 1 billion Base transfers and 200 to 250 million on Arbitrum - OP Mainnet displayed $0.00, which is a price and not a cost - Base held $14.6B of value and Arbitrum $12.3B, stocks that ignore a fee change ![Two receipts of different sizes pinned to the same ledger page](https://img1.ladyww.cn/picture/Picture00442.jpg) Base charged $0.001 to $0.002 per transfer and Arbitrum One $0.004 to $0.005 in Ethereum.org’s L2 directory snapshot of 1 to 6 October 2026. That is a threefold gap in the price of the same action. Total revenue can be bought with a campaign. Revenue per transaction cannot, because the campaign has to pay the price again on every unit. ## Why totals are easy to move Incentive programmes pay people to transact. Every subsidised transaction still prints a fee line, and in most designs the fee is charged and then rebated, or the user is paid in a token the venue issues. A campaign therefore raises the total while it lowers the average. Revenue per transaction is a division, and the denominator grows faster than the numerator when a subsidy is doing the work. A venue reporting a record revenue month may be reporting a fee policy that no user would choose without the rebate. This is not fraud. It is the difference between a level and a rate. Level metrics respond to spending, and rate metrics respond to pricing. ## The same revenue, different user counts Take one million dollars of revenue. On Base at $0.0015 per transfer that is about 667 million transfers. On Arbitrum One at $0.0045 it is about 222 million. Same money, three times the user count. Reverse the calculation and it becomes a screen. If two venues report the same revenue and one charges a third of the fee, the cheaper one either moved three times as many transactions or earned its revenue somewhere other than transfers. The published ranges widen the spread. At Base’s $0.002 ceiling, a million dollars is 500 million transfers. At Arbitrum’s $0.005 ceiling it is 200 million. That is the metric’s job. It converts a revenue headline into a count, and a count can be checked against blocks and against the venue’s own fee page. ## The fee is policy, not cost OP Mainnet displayed $0.00 in the same snapshot. Zero is a price, not a cost. The data cost of a rollup is decoupled from mainnet gas by blobs, so the marginal cost of a transfer is small and the posted fee is a decision about market share. The directory is therefore a map of pricing strategies rather than a cost table. Scroll sits at $0.002 to $0.003, Starknet at $0.005 to $0.008 paid in STRK, and Linea at about $0.024. Linea’s floor is roughly 12 to 24 times Base’s, for the same kind of transfer in the same week. Total value locked sees none of this. Base held $14.6 billion and Arbitrum One $12.3 billion. Both are stocks, and a stock barely moves when the price of a unit falls by a tenth of a cent. That is the argument in [L2 fee revenue, not TVL](/l2-fee-revenue-not-tvl/), and the per-transaction figure is its sharpest form. The same logic explains why a [sequencer that can reorder or delay](/what-a-sequencer-does/) a transaction is a different product at the same price. ## What the number cannot tell you The directory quotes a simple transfer. A swap or a bridge burns more gas and carries a different fee, so per-transaction revenue computed from a simple-transfer price is a floor for the blended network figure, not a measurement of it. The figures also sit near the reporting floor. $0.001 and $0.005 differ by five times, and both round to fractions of a cent in a headline. Precision at that scale needs the raw fee, not the rounded quote. And none of it describes duration. Base and OP Mainnet both take seven days to withdraw to mainnet. Arbitrum’s dispute window is 45,818 blocks, about 6.4 days. ZKsync Era advertises a minimum of three hours. A cheap transaction on a seven-day rail is cheap in the wrong dimension when the position has to move. | Network | Fee per transfer, 1 to 6 October 2026 | Transfers needed for $1M of revenue | |—|—|—| | Base | $0.001 to $0.002 | 500 million to 1 billion | | Arbitrum One | $0.004 to $0.005 | 200 to 250 million | | OP Mainnet | $0.00 displayed | Not defined | | Scroll | $0.002 to $0.003 | 333 to 500 million | | Starknet | $0.005 to $0.008, paid in STRK | 125 to 200 million | | Linea | About $0.024 | About 42 million | ## Verdict Revenue per transaction survives an incentive campaign because it is the one number the campaign cannot pay down without paying it again on every unit. It also inverts the usual ranking. On the 1 to 6 October 2026 snapshot, a million dollars of revenue on Base implies 500 million to 1 billion transfers and on Arbitrum One 200 to 250 million. That gap is not a quality judgment. It is a statement about fee policy, and fee policy is what a venue controls when it decides whether to rent activity or earn it. Ask for transfers alongside revenue, divide one by the other, and compare the answer with the fee on the page. If the campaign is doing the work, the rate will fall while the level rises, and only one of those two numbers is being paid for by users. ### FAQ Q: Why does revenue per transaction resist incentive campaigns? A: An incentive pays per unit of activity, so a subsidised transaction still books a fee line and still raises total revenue. The average falls because the subsidy, not the price, is generating the extra units. A campaign can therefore produce a record revenue month and a declining revenue per transaction at the same time. Q: How big is the gap between Base and Arbitrum fees? A: Roughly threefold in Ethereum.org's directory snapshot of 1 to 6 October 2026, with Base at $0.001 to $0.002 per transfer and Arbitrum One at $0.004 to $0.005. For $1 million of revenue that is 500 million to 1 billion transfers on Base and 200 to 250 million on Arbitrum One. Q: Does a lower fee mean a better network? A: No. The fee reflects pricing policy as much as cost, because blob data fees decoupled rollup costs from mainnet gas. A sub-cent fee can sit on a rail with a seven-day withdrawal, such as Base or OP Mainnet, or on one with a shorter exit. Read the fee with the exit time and the value held, not on its own. ## Liquid staking tokens: the depeg nobody models URL: https://www.caijing188.com/lst-depeg-risk/ Updated: 2026-10-27T22:09:43+08:00 Verified: 2026-10-27 ### Quick answer A liquid staking token converts the validator exit queue into a secondary-market price. Ethereum's exit queue held 850,000 ETH at about 14.8 days in October 2026, while Lido's 10% fee on a 2.63% gross leaves 2.37%, as of October 2026. Price the worse of the two exits. ### Key takeaways - The Ethereum exit queue held 850,000 ETH at about 14.8 days in October 2026 - The entry queue was roughly 2.5 million ETH, more than 43 days, on 22 July 2026 - A 10% fee on a 2.63% gross leaves 2.37%, so a 2.75% net implies a higher gross - More than 96% of ETH staking rewards came from issuance per Bitwise Q4 2026 figures ![A liquid staking token quoted below the value of the ETH behind it](https://img1.ladyww.cn/picture/Picture00432.jpg) A liquid staking token is a receipt for staked ETH, and the word liquid carries most of the argument. It does not remove the validator exit queue. It converts the queue into a secondary-market price, set by whoever will take the other side of your exit. Depeg models usually test that price against redemption value, which is the comparison that tells you least. ## What you hold, and what you can do with it Two routes exist for getting the ETH back. Redeem through the issuer or protocol, which places you in the validator exit queue and pays at the protocol conversion rate. Or sell the token, which pays immediately at whatever the book shows. The first route has a known rate and an unknown date. The second has a known date and an unknown rate. A peg assumes the two price the same thing at the same moment, and they do not. The gap between them is the risk. ## The queue is the real term sheet Ethereum’s exit queue held 850,000 ETH in October 2026, a wait of about 14.8 days. Three months earlier the mechanism ran the other way: on 22 July 2026 the entry queue held roughly 2.5 million ETH, a wait of more than 43 days. The queue is not a constant. It is a function of how many holders want the same exit at once, so it lengthens when confidence is weakest, and any model fixing the redemption delay has modelled the calm case. Total stake stood at 43.8 million ETH on 23 October 2026, which puts the October queue near 2% of all staked ETH. ## The fee line, and a number that does not reconcile The gross ETH staking rate in Q3 2026 was 2.63%, within a full-year range of 2.6% to 3.3%. Lido charges 10% of rewards. The board carries the net near 2.75%, which sits above the gross printed next to it. That only reconciles if the net was computed on a gross nearer the top of the band: 10% of about 3.06% leaves about 2.75%. Apply the same 10% to a 2.63% gross and the result is 2.37%. Read gross and fee on the same page before comparing wrappers. Centralised staking repeats the pattern: Coinbase charges about 35%, leaving 1.95% to 2.1%, Binance about 39.95%, leaving roughly 2.1%, and Kraken 10% to 26% by tier. More than 96% of ETH staking rewards came from new issuance rather than fees, per Bitwise’s Q4 2026 figures, so the yield is a dilution claim ([staking yield is not income](/staking-yield-is-not-income/)). ## The discount arrives when the queue is longest The secondary-market price is not a valuation of ETH. It is the price of skipping the queue. In calm conditions the queue is short, so skipping it is worth almost nothing and the token trades near redemption value. In stress the queue is long, skipping becomes valuable, and pricing it means a seller accepting a discount against a buyer who must hold through the same queue. A discount is a fee for immediacy, and the fee rises when immediacy is what you want. Book depth thins at the same moment, because makers are warehousing the same duration mismatch. The token also travels into lending markets, where the discount becomes a health factor, marked down in the same week the borrow rate resets ([how utilisation sets lending rates](/lending-rates-utilization/)). ## A peg that is a spread A liquid staking token has no peg. It has a spread between two exits. | Exit route | Time to ETH | Price certainty | Fails when | |—|—|—|—| | Redeem through protocol | About 14.8 days at October 2026 queue depth | High, set by conversion rate | The queue lengthens further | | Sell on the secondary market | Immediate | Low, set by the book | Depth thins and the discount widens | | Hold the token | Indefinite | Marked to redemption | Never, but you stay exposed | | Run a validator | Queue plus setup | High | Needs 32 ETH and operations | A depeg model built on redemption value reads one row and ignores the other three, and the row that matters in a drawdown is the second. ## What to do next Size the position on the exit you would use in a drawdown, not the one on the product page. If your plan is to sell, check book depth at your size and treat the stressed-week discount as a cost, not a dislocation to capture. Before comparing two wrappers on yield, restate both nets from the same gross with the fee applied in front of you. The gap between 2.75% and 2.37% is small, and it decides which wrapper an allocator picks. ## Verdict A liquid staking token trades withdrawal timing for market pricing. In October 2026 that timing was about 14.8 days on an 850,000 ETH queue, and the market price is the only one that moves before the queue clears. Treat the token as a claim with two exits and price the worse one. The peg is not a fact about the asset. It is a spread, and it widens when you need it narrow. ### FAQ Q: Does a depeg mean the underlying ETH is impaired? A: No. A discount on a liquid staking token reflects the cost of skipping the withdrawal queue, not a loss of the staked ETH. The redemption rate through the protocol is unchanged. The discount is what you pay a buyer who accepts the queue in your place. Q: Why is the discount worst when the queue is longest? A: Because the queue and the discount are two prices for the same thing, which is immediacy. When the queue is short, waiting is cheap and the token trades near redemption value. When the queue lengthens, buyers demand compensation for the wait and sellers who need cash accept it. Q: How should two liquid staking tokens be compared on yield? A: Recompute both nets from the same gross rate at the same timestamp, with the fee applied in front of you. Published net figures can be computed on a different gross rate from the one printed beside them, which is how a 10% fee can appear to raise a yield. ## Order book depth is the number your slippage model is missing URL: https://www.caijing188.com/depth-over-volume/ Updated: 2026-10-27T22:09:44+08:00 Verified: 2026-10-27 ### Quick answer 24h volume measures turnover, not resting size, and a venue can rank high on our board of 24 venues as of October 2026 while staying thin inside one percent. FeeEdge's 17 August 2026 survey puts Solana withdrawals near 0.5 to 1 USDT. Read depth at your size minutes before the order, not the day before. ### Key takeaways - Volume is a historical aggregate; depth is a snapshot of orders you can actually hit - The board ranks 24 venues on sampled 24h volume as of October 2026, and the top row carries about a third of that sample - The ratio of 1% to 2% depth separates a wall from a cliff behind it - FeeEdge's 17 August 2026 survey priced Solana SPL withdrawals at roughly 0.5 to 1 USDT ![An order book ladder next to a 24 hour volume bar chart](https://img1.ladyww.cn/picture/Picture00402.jpg) Volume tells you what traded. Depth tells you what you can trade. A slippage model that takes the first number and assumes the second will underprice every large order, and it will underprice it worst on the venues that look most liquid on a ranking page. The gap between the two numbers is where execution costs hide. ## Two numbers, two questions 24h volume answers a historical question: how much changed hands in the last day. Depth answers a present-tense one: how much can change hands in the next minute without moving the mid by a given amount. Both are legitimate. They are not interchangeable. Volume is an aggregate over time and over both sides of the book. Depth is a snapshot of resting orders, and only orders that are actually resting count. A venue can score high on the first and low on the second for ordinary reasons. Flow may be concentrated in one pair with a small average ticket, makers may be obliged to quote only near the touch, and volume may come from participants who hedge elsewhere. ## Why a volume ranking can hide a thin book Every ranking has to sort on something, and volume is the easiest column to compute. Our board sorts the same way: it ranks 24 venues on sampled 24h volume as of October 2026. The sort key is a choice, and it shows in the shape of the table. The top row carries roughly a third of the sampled volume. Twenty-three venues share what remains, and the drop from the first row to the second is steeper than the drop from the second to the tenth. That shape is a warning rather than a measurement. It says the sample is dominated by one venue, so the median row describes a book much thinner than the table’s total implies. A trader who sizes to the total is sizing to a number that no single book ever held. We do not publish per-venue depth, and the board does not claim to. What it can do is stop the volume column from being read as a liquidity column. ## Reading depth inside one percent ### The ratio between two bands Depth inside 1% is the notional you can execute before the mid moves one percent. Depth inside 2% is the same test at a wider band. The ratio between them says how fast the book thins. A book with a large figure inside 1% and almost the same figure inside 2% is a wall with a cliff behind it. A book with a small figure inside 1% and a far larger one inside 2% is thin at the touch and deep behind it. The second shape needs a different execution style from the first. ### Where estimates go wrong Most retail slippage models assume depth is proportional to volume and constant through the day. Neither holds. Depth clusters around the moments when makers refresh, and it thins in the minutes around a large liquidation or a funding timestamp. A second error is averaging. Slippage is convex in size, so doubling an order more than doubles the cost, because the second half eats worse prices than the first. A model built on one average figure is optimistic exactly where it matters. ## What the board can and cannot tell you The board can tell you which venues are in the sample, how the sample was sorted, and when it was taken. It cannot tell you the depth of a book it did not snapshot. That limit is worth stating plainly, because a ranking invites the opposite reading. A row is a position in a sort, not a certification of liquidity. The [methodology](/methodology/) page sets out what the sample covers and what it leaves out. | Measure | What it answers | Failure mode | |—|—|—| | 24h volume | How much traded inside the window | Read as resting size | | Depth inside 1% | Size tradeable before a one percent move | Snapshot goes stale within minutes | | Ratio of 1% to 2% depth | How fast the book thins | Hides a cliff behind a wall | | Maker count at the touch | Whether depth will be replaced | One desk can run several quotes | | Published withdrawal fee | Cost of the exit | Changes without notice | ## What to do next Pull the book at the size you intend to trade, minutes before the order rather than the day before. Compare depth inside 1% with depth inside 2%, and count makers at the touch to judge whether that depth survives your order. Then price the exit. A position is only as liquid as the way out, and the exit cost is published. FeeEdge’s survey of 17 August 2026 put exchange withdrawal fees at roughly 0.5 to 1 USDT for Solana SPL, about 0.3 USDT for BEP-20, and highest for ERC-20, with L1 ERC-20 transfers at $2 to $10 before congestion. Use the [venue directory](/exchange/) to see how many books an order can be split across. Treat the volume ranking as the start of that comparison, not the end of it. The number a slippage model is missing is the one that describes the book at the moment the order crosses it. ### FAQ Q: Can a venue have high volume and thin depth at the same time? A: Yes, and it is common. Volume counts turnover across a day and both sides of each trade, including makers trading with each other. Depth counts only orders resting at this moment. A venue whose flow is concentrated in small tickets, or whose makers quote only at the touch, will look liquid on volume and thin on depth. Q: What does depth inside one percent actually measure? A: It is the notional you can execute before the mid price moves one percent in either direction. Comparing it with depth inside two percent shows how quickly the book thins, which is the part a single average slippage figure hides. Q: Why is averaging slippage across sizes misleading? A: Slippage is convex in size. Doubling an order more than doubles the cost, because the second half fills at worse prices than the first. A model built on one average figure is therefore optimistic at exactly the order sizes where the estimate matters most. ## Stablecoin rules did not kill yield, they moved it URL: https://www.caijing188.com/stablecoin-yield-rules/ Updated: 2026-10-27T22:09:44+08:00 Verified: 2026-10-27 ### Quick answer Rules that stop a stablecoin issuer paying a return push the return into a separate product, as of October 2026, when tokenised Treasury bills yielded 3.60% to 3.61% on a seven-day basis (RWA.xyz, 7 October 2026). Total stablecoin supply was $292.40B that month. Check what you hold, not what the page calls it. ### Key takeaways - Tokenised Treasury bills paid 3.60% to 3.61% annualised on 7 October 2026, the benchmark for a dollar balance - Tokenised Treasury products held $14.64B to $14.90B across 117 funds in October 2026 (RWA.xyz) - A 6% rewards rate on a stablecoin is a different instrument, not a better version of the same one - Total stablecoin supply was $292.40B in October 2026, and where it sits decides what it earns ![A stablecoin balance beside a fund factsheet and a yield chart](https://img1.ladyww.cn/picture/Picture00422.jpg) A rule that stops a stablecoin issuer paying you a return does not delete the return. It relocates it, usually into a second product with a second set of documents. The number that used to sit on the balance is now a yield on a fund unit, a lending position or a programme called rewards. The reader’s job is to notice which of those they are actually holding. ## What the rule restricts Frameworks that treat a stablecoin as a means of payment rather than a deposit commonly restrict the issuer from paying a return on the outstanding balance. The reasoning is that a payment token promising interest starts to look like a deposit, and deposit-taking is a licensed activity with a guarantee attached. The restriction usually lands on the issuer and on the reserve. It does not ban yield in the ecosystem. It bans the simplest version of it, where the balance itself accrues. That distinction explains most of the product shelf. The token stays boring, and the yield moves one layer out. ## Where the return went ### The tokenised fund The cleanest replacement is a fund unit holding short government debt. The holder owns a share of a portfolio rather than a claim on an issuer’s balance, and the yield is the portfolio’s yield net of fees. Tokenised Treasury products held $14.64B to $14.90B across 117 funds in October 2026 (RWA.xyz), and the seven-day annualised rate was 3.60% to 3.61% on 7 October 2026. ### The lending venue The second route is a lending market, where the return is a borrowing rate rather than an issuance policy. That rate moves with utilisation, which is why it can exceed a T-bill and can also fall to almost nothing when a pool is idle. Our note on [lending rates and utilisation](/lending-rates-utilization/) walks through the mechanism. ### The rewards programme The third route keeps the word stablecoin and adds a programme. Rewards may be paid in the issuer’s own token, in points, or as a rate that applies only to a promotional balance. The economics may be identical to interest. The disclosure is usually thinner, and that is the grey zone. ## The 3.60% benchmark 3.60% to 3.61% is now the number to beat. It is a short-dated government yield, published with a date, reachable through a product that discloses its portfolio. A stablecoin programme offering more is offering something else alongside the dollar. That comparison is the whole argument. A rewards rate well above the benchmark on a dollar balance is not automatically better than 3.61% on a fund unit. It is a different instrument with a different failure mode, and the page that advertises it rarely puts the two side by side. The same discipline applies to the stablecoin itself. Total supply was $292.40B in October 2026, and [who holds it](/stablecoin-supply-who-holds-it/) is a separate question from what it earns. ## The grey zone is the disclosure Ask three questions of any rewards programme. What funds the payment: reserve income forgone, a token subsidy, or lending revenue? Who is the counterparty if the programme stops? And is the rate contractual or discretionary? A programme that cannot answer the first question is being funded from somewhere you cannot see. That is not automatically unsafe. It is unpriceable, which for a treasury position is worse. Compare net of fees, too. For instance, a 3.60% gross yield with half a point of fees nets 3.10%, and we check fee pages monthly because the gross number is the one that gets advertised. | Where the return sits | What you actually hold | Rate and period | |—|—|—| | Tokenised Treasury fund | A fund unit backed by short government debt | 3.60% to 3.61% annualised, seven-day (RWA.xyz, 7 October 2026) | | Ethereum staking | A claim on validator rewards | 2.63% gross in Q3 2026 (staking data, Q3 2026) | | Solana delegation | A stake delegated to a validator | 5.99% gross, 4.6% to 7.6% net after commission (Solana staking data, Q3 2026) | | Lending market | A claim on borrowers | Variable, set by pool utilisation | | Rewards programme | Usually a promise, not an asset | Discretionary, and often undisclosed | ## What to do next Divide the balance into what is collateralised, what is promised and what is insured. The stablecoin is a claim on an issuer. The fund unit is a claim on a portfolio. The rewards programme is a promise, and it should be sized like one. Then read the fee page before the marketing page, because the number that matters is net. If the programme will not tell you what funds it, treat the rate as zero for planning purposes and be pleasantly surprised. The relocation is the story. Yield did not disappear from the dollar. It moved to products that ask you for identity documents and disclose a portfolio, and that is a fair trade to understand before you make it. ### FAQ Q: Why would a rule stop a stablecoin issuer paying interest? A: Because a payment token that accrues a return starts to resemble a deposit, and deposit-taking is normally a licensed activity with protection attached. Frameworks that classify a stablecoin as a means of payment therefore commonly restrict the issuer from paying a return on the outstanding balance. The restriction usually lands on the issuer and the reserve, not on the whole market. Q: Is a rewards programme the same as interest? A: Economically it can be, but the disclosure is different. Interest is a contractual rate on a balance. Rewards may be paid in the issuer's own token, in points, or only on a promotional balance, and the source of the funding is often unstated. Ask whether the rate is contractual or discretionary before treating it as yield. Q: What should I compare a stablecoin yield against? A: A short-dated government yield available through a disclosing product. Tokenised Treasury bills at 3.60% to 3.61% annualised on 7 October 2026 are the obvious reference. Anything materially above that is offering credit risk, duration risk, token risk or a subsidy that ends. ## Finality is a budget, not a switch URL: https://www.caijing188.com/finality-as-a-budget/ Updated: 2026-10-27T22:09:45+08:00 Verified: 2026-10-27 ### Quick answer Finality is a trade of time for certainty, not a binary state. Ethereum.org's directory, as of October 2026, lists 7-day exits on Base and OP Mainnet, a 45,818-block dispute window on Arbitrum, about 6.4 days, and a 3-hour minimum on ZKsync Era. An exchange credits on a confirmation count; a bridge waits for the window. ### Key takeaways - Ethereum.org's October 2026 directory puts Base and OP Mainnet exits at 7 days - Arbitrum's dispute window is 45,818 blocks, about 6.4 days, in the same listing - ZKsync Era quotes a 3-hour minimum on Ethereum.org's October 2026 listing - A confirmation count and a challenge window answer different questions, so they are not comparable ![Two clocks set to different finality windows](https://img1.ladyww.cn/picture/Picture00412.jpg) Finality is usually described as a property a chain has. It reads better as a budget a chain spends. Every design buys a kind of certainty with a fixed amount of time, and the price of that purchase is the number an exchange desk, a bridge and a user each experience differently. ## Two kinds of certainty Probabilistic finality is the older model. A block gets harder to reverse with every block built on top of it, and the cost of reversal rises with each confirmation. Nothing is ever finished. The curve simply gets steep enough that no rational attacker follows it. Economic finality makes the same argument explicit. Reverting settled state would require burning capital that exceeds the value of the reversion, so the guarantee is a price rather than a timestamp. A chain is final when attacking it costs more than the attack is worth. Proof-based finality is the third case, and the one people usually mean when they say a chain is final. A validity proof or a passed challenge window converts a statement that is merely expensive to reverse into one that cannot be reversed without breaking a rule the verifier checks. It is the strongest of the three claims, and it is always paid for in latency. ## Why a deposit desk counts confirmations An exchange crediting a deposit is not verifying finality. It is running a risk model. The desk picks a confirmation count, watches for reorgs deeper than that count, and treats the residual risk as a cost of doing business. The credit appears fast because the venue, not the chain, is carrying the exposure. That is a rational trade for a venue with a treasury and a fraud team. It is a poor trade for anyone who reads the credit as proof. Between the credit and the settlement there is a window in which your balance is an entry in someone else’s database, backed by their solvency rather than by a state root. The count itself is a policy number, set per chain and revised when reorg behaviour changes. It is not published as a guarantee, and it should not be read as one. ## Why a bridge waits A bridge moves a claim rather than a balance, so it cannot settle on a risk model it controls. It has to wait until the claim is enforceable, which on an optimistic rollup means waiting out the challenge window in full. That is why the clock is measured in days, and why the number differs by design. Ethereum.org’s L2 directory, in its October 2026 listing, shows Base and OP Mainnet at 7 days to mainnet, Arbitrum’s dispute window at 45,818 blocks, about 6.4 days, and ZKsync Era at a 3-hour minimum. The first two are waiting for the right to object to expire. The third is waiting for a proof to be produced and checked. Neither is a defect. Both are the published price of a settlement model, and a user who moves size without reading that price is guessing at the risk. | Route | Exit to mainnet, Ethereum.org, October 2026 | What the clock actually buys | |—|—|—| | Base | 7 days | Expiry of the challenge right on an optimistic rollup | | OP Mainnet | 7 days | The same model under a different operator | | Arbitrum One | 45,818 blocks, about 6.4 days | A dispute window counted in blocks, not days | | ZKsync Era | 3 hours minimum | Proving and verification latency | | Centralised venue | No chain window, venue policy decides | Custodial credit on the venue’s book | ## What the window costs you A finality window is not idle time. It is time in which your capital is committed, your price is not, and your optionality is gone. Send the same notional through a 7-day exit and a 3-hour exit, the windows Ethereum.org listed in October 2026, and you are not comparing fees. You are comparing how long a position stays unreachable. On a volatile asset, a week of unreachable capital is a directional bet you did not choose to make. The fee side of the trade is small by comparison. FeeEdge’s survey of 17 August 2026 put L1 ERC-20 withdrawals at $2 to $10 and above $50 under congestion, with Base transfers at under $0.01 to $0.10 and Arbitrum at $0.01 to $0.50. Those numbers are visible before you commit, though [fee pages change](/fee-pages-change-monthly/) more often than most traders assume. The window is not priced anywhere, which is exactly why it gets ignored. ## Verdict Treat finality as a budget line, not a checkbox. Ask what the chain spends to give you certainty, and how long your capital sits inside that spend. An exchange credit is a fast, cheap promise from a counterparty. A bridge exit is a slow, expensive proof from a protocol. Both are legitimate, and the mistake is using one and assuming the other. Before you move size, read the exit window next to the fee, and check [who secures a rollup](/who-secures-a-rollup/) to see which party your window depends on. If a route cannot tell you its clock, it has told you something already. ### FAQ Q: Is a transaction final when the exchange credits my deposit? A: No. The credit is the exchange's own promise, issued after a confirmation count it chose, and it is backed by the venue's balance sheet rather than by the chain. The chain's own settlement arrives later, which is why a deep reorg is an exchange problem before it is a user problem, as of October 2026. Q: Why does a bridge take days when a deposit takes minutes? A: A bridge moves a claim, not a balance on someone's ledger, so it has to wait until that claim is enforceable. On an optimistic rollup that means waiting out the challenge window. Ethereum.org's October 2026 listing shows 7 days for Base and OP Mainnet and 45,818 blocks, about 6.4 days, for Arbitrum. Q: Does a shorter exit mean a weaker guarantee? A: Not necessarily, but it means a different mechanism. ZKsync Era's 3-hour minimum, per Ethereum.org in October 2026, comes from proving state rather than waiting out a challenge. What changes is what you trust: a proof that is checked, or a window in which someone is allowed to object. ## Seed backups that survive a fire, and the one that does not URL: https://www.caijing188.com/seed-backup-that-survives/ Updated: 2026-10-27T22:09:46+08:00 Verified: 2026-10-27 ### Quick answer A seed in one location loses to fire, water or burglary, and a photograph loses to camera roll sync. The stablecoin float was $292.40B as of October 2026, per this board's data pack, and much of it sits behind one piece of paper. Split the backup across locations, keep it offline, and test the restore while the balance is small. ### Key takeaways - Fire and water are one failure mode: they destroy the location, not the format - A photo of a seed joins the cloud backup, the shared album and the old phone - A metal plate survives the fire and tells any informed visitor what it is - A test restore costs one withdrawal, $2 to $10 for an ERC-20 token on Ethereum L1 in FeeEdge's 17 August 2026 sample ![Stamped metal seed plate beside a paper backup](https://img1.ladyww.cn/picture/Picture00452.jpg) A seed backup is a bearer instrument. Whoever finds it owns the funds. Most advice stops at “write it down and keep it safe”, which is the part that fails first: one location loses to fire, water and burglary, and the two convenient upgrades, a photograph and a metal plate, each fail in their own way. The design that holds is split, offline, and tested before it protects a real balance. ## The failures, in the order they happen ### Fire and water A house fire and a burst pipe are one event for a paper backup. Both destroy a location and neither leaves time to retrieve anything. A safe rated for documents helps with fire and not with flood, and few read the rating before buying. ### The burglar takes the container The second failure is quieter. A thief who takes the safe, the drawer or the filing box takes the backup, and a seed needs no key to spend. Storing it next to the device is the version everyone believes they are too clever to do. ## The photograph is the one that fails silently Photographing a seed feels like redundancy, and technically it is. The problem is where the photograph goes. Camera rolls sync to a cloud account by default, within seconds of the shot. From there the image joins the account backup, the family sharing group, the desktop photo library and whatever third party app was granted library access. Nothing about this is visible. There is no notification when a picture of 24 words is uploaded, indexed and retained in a version history. Account takeover, a shared album link, a forgotten old phone: any of those now includes the seed. It works until it is the reason funds are gone, which makes it the worst of the set. ## Metal fixes durability, not secrecy A stamped steel plate survives fire, flood and a decade in a drawer. It does not survive a visitor who knows what 24 stamped words mean. Metal plates are recognisable, often labelled, and frequently sold with a holder that announces its purpose. Two adjustments help. Leave the plate unlabelled and store it somewhere that is not obviously a safe, and treat the plate as one share rather than the whole secret. Durability is a property of the material. Secrecy is a property of the layout. ## What splitting actually buys Splitting turns one object into several, and the value depends on how you split it. ### Shamir shares A 2-of-3 scheme generates three shares, any two of which rebuild the seed. No single location holds a secret, a burglar who finds one share has nothing usable, and losing one location is survivable. The cost is tooling: a generator you trust, a documented threshold, and a record of which share is where. ### Halves of a word list Splitting 24 words into two sets of 12 is simpler and needs no software. One half alone does not reconstruct the wallet, so a single theft is not fatal. The failure is correlation: lose either half and the wallet is gone, so the scheme has the same availability as a single copy while doubling the number of places that must survive. | Method | Survives a fire at one site | Survives a thief who knows crypto | New way to lose everything | |—|—|—|—| | Paper in a drawer | No | No | Any single event at that address | | Photo in the camera roll | Yes | No | Cloud account takeover | | Encrypted file in email | Yes | Partly | Forgotten password, provider lockout | | Metal plate at home | Yes | No | A visitor who reads it | | Two halves, two addresses | Partly | Partly, one theft only | Losing either half | | Shamir 2-of-3, three addresses | Yes | Yes, below two shares | Losing two shares | ## Test once, before the size is real An untested backup is a hypothesis, and the test belongs before the balance matters. Restore the seed on a spare device or an offline software wallet, confirm the derived address matches the funded one, then wipe it. Send a small amount, spend it, restore again. That proves the words, the derivation path and your ability to follow the procedure. The test is cheap next to the position. Moving an ERC-20 token on Ethereum L1 cost $2 to $10 in FeeEdge’s 17 August 2026 sample, and the same transfer on Base costs a fraction of a cent per Ethereum.org’s 1 to 6 October 2026 snapshot. The stablecoin float was $292.40B in October 2026, per this board’s data pack, and much of [that supply](/stablecoin-supply-who-holds-it/) sits behind one piece of paper. ## What to do next Write the location map, not the words. The map says which share is in which city and who else knows the procedure. It never contains a word of the seed. Pick the number of locations from the size, not from enthusiasm. Below a few months of expenses, one hidden offline copy is proportionate. Above that, split, and keep the [savings balance](/how-much-in-a-hot-wallet/) smaller than the backup procedure can carry. Then put a date in the calendar, once a year, to confirm every location still exists. Houses are sold and safes are moved. ### FAQ Q: Is a photo of my seed safe if my cloud account has two-factor login? A: No. The photo joins a library that syncs, indexes and retains versions, and any account takeover, shared album link or forgotten device includes it. Two-factor protects the account, not the copies already stored inside it. Q: Is splitting 24 words into two halves of 12 safe? A: It removes the single object and stops a thief who finds one half, but it keeps the availability of a single copy: lose either half and the wallet is gone. Shamir sharing with a 2-of-3 threshold tolerates one lost location as well as one theft, at the cost of tooling and a documented procedure. Q: How often should I re-test a backup? A: Once a year is a reasonable default, and after any move, renovation, device change or change of address. The test is a restore on a spare device with the derived address compared against the funded one, followed by a small spend. Locations that no longer exist are not backups. ## When a stablecoin’s $292B float is really three balance sheets URL: https://www.caijing188.com/float-concentration/ Updated: 2026-10-27T22:09:46+08:00 Verified: 2026-10-27 ### Quick answer Stablecoin supply reached $292.40 billion in October 2026, but that is a stock, not a measure of who holds it. Issuer share and chain share describe rules and routes, as of October 2026. Only holder concentration predicts how fast a stressed float can exit. ### Key takeaways - Total stablecoin supply was $292.40 billion in October 2026, a stock rather than a flow - Issuer share explains whose redemption window and freeze list apply - Chain share explains cost: ERC-20 at $2 to $10 and Tron at $1 to $3 on 17 August 2026 - Holder concentration is the only one of the three that predicts a stress exit ![Three separate ledgers being balanced against one another](https://img1.ladyww.cn/picture/Picture00443.jpg) Stablecoin supply stood at $292.40 billion in October 2026. That figure is a stock, and a stock says how much exists rather than how much can leave. A float concentrated in a few desks behaves like inventory. It does not evaporate under stress, it moves, and the speed of the move is set by how few people have to agree. ## A stock is a balance, not a flow Every stablecoin headline is a balance. It counts tokens outstanding at a moment. It does not count holders, transfers per day, or the terms on which an issuer will take a token back. Three different measures get quoted as if they were one. Issuer share splits the float by who mints it. Chain share splits it by where the token sits. Holder concentration splits it by who controls it. Only the third is a statement about behaviour, and behaviour is what a stress event tests. ## Measure one: issuer share Issuer share answers a legal question. It tells you whose redemption window, whose freeze list, and whose attestation cadence apply to a given block of supply. That is useful and it is not a stress measure. A large issuer can have a dispersed holder base, in which case redemptions arrive as a slow trickle of small tickets. A smaller issuer can be one desk’s treasury, in which case a single decision removes most of its float. The share tells you who administers the exit, not how many people are walking toward it. ## Measure two: chain share Chain share answers a routing question. It tells you which settlement layer absorbs the traffic when holders move, and what that move costs. The cost is published. FeeEdge’s survey of 17 August 2026 put ERC-20 transfers on Ethereum L1 at $2 to $10, with congestion above $50, Tron at $1 to $3, and BNB Chain at $0.01 to $0.20. A chain split therefore predicts the fee a stressed holder pays and the rail they queue on. It does not predict crowding. Tokens can sit on a chain for years untouched, and a large chain share is often a large custody share rather than a large demand share. ### Why the chain table gets misread Exchange wallets and bridges hold tokens for people who have never moved them, on whichever chain the venue chose. That is an engineering and liquidity decision, not a signal about who intends to sell. ## Measure three: holder concentration Holder concentration answers the only question that matters in a stress event. How many independent decisions stand between the float and a redemption order. Redemption is an action taken by an owner. If ten desks control most of a $292.40 billion float, the float is not a market. It is inventory with ten custodians. The headline supply does not change when one of them decides to leave, but the redemption queue does, and so does the price of the asset it is swapped into. This measure also has the worst data. Address labels are an inference. One desk can run a treasury address, several market-maker addresses, and a custody account, so raw address counts overstate dispersion. | Measure | What it answers | What it misses | Predicts a stress exit | |—|—|—|—| | Total supply | How many tokens exist | Who holds them, on what terms | No | | Issuer share | Whose rules and redemption window apply | Whether the holders are one desk | No | | Chain share | Which rail and fee absorb the flow | Whether anyone intends to move | No | | Holder concentration | How many decisions precede a redemption | Whether the labels are right | Yes | | Address count | Apparent dispersion | One desk, many addresses | Weakly, and backwards | ## What the board can and cannot show The board shows supply, and that number is dependable because anyone can recompute it from the chain. Holder tables are not in the same category. They depend on labels maintained by third parties, with their own errors. That distinction is the argument in [who actually holds the stablecoin float](/stablecoin-supply-who-holds-it/). The same logic applies to reserves. An attestation is a snapshot of what backs the tokens at a timestamp, not a guarantee of what those reserves will be worth next week, which is why [proof of reserves is three numbers, not one](/proof-of-reserves-three-numbers/). So the honest reading of $292.40 billion is narrow. This much exists. We do not know from the headline how much of it is one phone call away from the exit. ## Verdict Supply is the least informative of the three measures and the only one that gets quoted. Issuer share tells you whose rules apply. Chain share tells you what the exit costs, with ERC-20 transfers at $2 to $10 and Tron at $1 to $3 in the 17 August 2026 survey. Holder concentration is the one that predicts how a stress event resolves, because redemptions are decisions and a concentrated float concentrates decisions. A desk that wants to know whether a $292.40 billion market can absorb a bad week should ask for the holder table, then treat its absence as part of the answer. The headline is a size. The question is a queue. ### FAQ Q: What are the three ways to measure stablecoin concentration? A: Issuer share splits the float by who mints it and therefore whose redemption rules and freeze powers apply. Chain share splits it by where the tokens sit, which sets the fee and the rail used to move. Holder concentration splits it by who controls the tokens, which is the only measure that says how many independent redemption decisions exist. Q: Why does holder concentration predict a stress exit? A: A redemption is an action taken by an owner, so the number of owners behind the float sets how many decisions stand between the market and a queue. With ten desks controlling most of a $292.40 billion float, one decision can move a large share of supply while the headline stays unchanged. Issuer and chain shares say nothing about who decides. Q: Can address counts substitute for holder concentration? A: No, and they usually point the wrong way. One desk can operate a treasury address, several market-maker addresses, and a custody account, so raw address counts overstate how dispersed the float really is. A holder table built on addresses will look more distributed than the behaviour it describes, which is why labels matter more than counts. ## Collateral you cannot liquidate is not collateral URL: https://www.caijing188.com/illiquid-collateral/ Updated: 2026-10-27T22:09:47+08:00 Verified: 2026-10-27 ### Quick answer Tokenised Treasuries carry almost no price risk and a great deal of time risk, because most redeem only through the issuer in business hours after KYC. DefiLlama put on-chain AUM at $34.865 billion against $4.182 billion deployed in DeFi in October 2026, as of October 2026. The gap is liquidity, not demand. ### Key takeaways - DefiLlama put on-chain AUM at $34.865 billion against $4.182 billion deployed in October 2026 - RWA.xyz recorded $39.01 billion of distributed value on 23 October 2026, up about 81% this year - Morpho Blue held $735.9 million and Aave V3 $437.6 million of the deployed slice - Tokenised real estate was about $226 million, under 1%, because redemptions take months ![A tokenised Treasury fund next to a liquidation engine](https://img1.ladyww.cn/picture/Picture00433.jpg) Collateral is not a valuation. It is an asset a lender can convert to cash on a schedule the lender controls, at a price that survives an actual liquidation. A tokenised Treasury bill passes the first test and can fail the second, and that gap explains most of the on-chain real-world asset book. ## The gap is liquidity, not demand DefiLlama put on-chain AUM for tokenised products at $34.865 billion in October 2026. The amount deployed into DeFi was $4.182 billion. That is about 12%. Demand is not the constraint. RWA.xyz recorded $39.01 billion of distributed value on 23 October 2026, up from roughly $21.5 billion at the start of the year, a rise of about 81%. Tokenised Treasuries accounted for $14.64 billion to $14.90 billion across 117 funds in the 16 to 23 October 2026 readings, and those products quoted 3.60% to 3.61% annualised on a seven-day basis on 7 October 2026. The assets exist, they pay something, and they are large. What they lack is a redemption path a liquidation engine can use on a Sunday night. ## A Treasury that redeems in business hours is a poor liquidation asset Most tokenised Treasury products require KYC to mint and to redeem. Redemption runs through the issuer, in the issuer’s banking hours, at the issuer’s settlement speed. Now post that asset as collateral and let the position go underwater on a Saturday. A liquidator has to seize the collateral, sell it, and repay the debt quickly enough that the position does not deteriorate before the bonus is collected. If the only exit is an issuer desk that opens on Monday, the liquidator is not buying collateral. They are buying a claim with a settlement date and a counterparty. The discount they demand for that is not credit risk, it is redemption risk, and it is charged when the market is least able to pay it. A Treasury has almost no price risk. It has time risk, and time is what a liquidation does not have. ## Why the dashboard says the collateral is fine Most lending dashboards show a loan-to-value derived from net asset value, fed by an oracle, with a haircut applied. The haircut prices volatility. It does not price the queue. The result is a number that looks conservative while measuring the wrong variable. A 5% haircut on a 3.6% yielding Treasury reads as prudence against a price move that will essentially never happen, while the real exposure, a multi-day settlement window spanning a weekend, appears nowhere on the page. | Asset | Redemption route | Time to cash | Works as liquidation collateral | |—|—|—|—| | Tokenised Treasury | Issuer, business hours, KYC | Days, not minutes | Only with a market maker obliged to bid | | Tokenised private credit | Issuer, fund terms | Weeks to quarters | No, absent a secondary desk | | Tokenised equity | Broker or issuer | Days | Rarely, books are thin | | Native stablecoin | On-chain, continuous | Minutes | Yes | | ETH | On-chain, continuous | Minutes | Yes, with a volatility haircut | Real estate sits at the far end. RWA.xyz put tokenised real estate at about $226 million in October 2026, under 1% of distributed value, because a token representing a building has a redemption route measured in months. ## Where the $4.182 billion actually goes The deployed slice is concentrated. Morpho Blue held $735.9 million of it and Aave V3 $437.6 million in DefiLlama’s October 2026 readings. Those are venues with professional liquidators, permissioned pool structures in some cases, and a market maker willing to bid. Tokenised Treasuries work as collateral where somebody has pre-arranged the exit, usually a market maker or a whitelisted pool that can absorb the asset against stablecoins without waiting for the issuer. The token is not the collateral. The agreement around the token is, which is the distinction that runs through [the three products carrying the RWA book](/thirty-nine-billion-three-products/). Collateral quality is a lending parameter, not a badge, and it changes the rate a market can offer ([how utilisation sets lending rates](/lending-rates-utilization/)). ## What to do next If you lend against tokenised real-world assets, ask three questions before reading the LTV. Who can redeem this at 3am on a weekend, without KYC? How long does that take, and what did it cost the last time the market moved fast? And what happens to the liquidation bonus if the only buyer is the issuer’s own desk? If you hold the token as a yield position rather than as collateral, the same questions apply to your exit. The 3.60% to 3.61% seven-day yield quoted on 7 October 2026 is a return on an asset whose liquidity you have lent to the issuer. ## Verdict The on-chain RWA trade was described as a demand problem for years. The numbers say otherwise: $39.01 billion of distributed value and $34.865 billion of on-chain AUM against $4.182 billion deployed in DeFi as of October 2026. What is missing is a redemption path that survives a liquidation. Collateral you cannot liquidate is not collateral. It is a marked position with a settlement date, and it should be sized like one. ### FAQ Q: Why can a tokenised Treasury not simply be sold on-chain? A: It can, if a market maker will bid. The problem is that the natural buyer is another whitelisted holder and the issuer's redemption desk is closed. In a fast market the only bid comes from someone who will carry the settlement delay, and they charge for it. Q: Is a haircut on net asset value enough protection for a lender? A: It protects against price movement, which is the smaller risk for a short-duration Treasury. It does not protect against the time needed to convert the token into cash. A haircut that ignores redemption time measures the wrong variable, however conservative it looks. Q: Where do tokenised Treasuries work as collateral today? A: In venues where the exit is pre-arranged, typically permissioned pools with professional liquidators and a market maker obliged to bid. Morpho Blue and Aave V3 held most of the $4.182 billion deployed in October 2026, a small fraction of the $34.865 billion on chain. ## Why two exchange rankings disagree about the same venue URL: https://www.caijing188.com/volume-rankings-disagree/ Updated: 2026-10-27T22:09:47+08:00 Verified: 2026-10-27 ### Quick answer Two rankings can disagree about one venue because pair universe, wash filters, quote currencies and sampling windows differ. Our board ranks 24 venues on sampled 24h volume as of October 2026, with the top row near a third of the sample. Read the methodology before quoting a rank. ### Key takeaways - Volume is an estimate produced by a filter, and the filter is part of the estimate - Perp inclusion is the largest single source of gap between otherwise similar rankings - Stablecoin supply reached $292.40B in October 2026 per RWA.xyz, so USD conversion touches nearly every pair - The board ranks 24 venues on sampled 24h volume as of October 2026, and the top row carries about a third ![Two exchange ranking tables showing the same venue in different positions](https://img1.ladyww.cn/picture/Picture00403.jpg) Two rankings, one venue, two different positions. The usual explanation is that one of them is wrong. The more common explanation is that they measured different things and both reported honestly. A volume ranking is a methodology before it is a fact, and the methodology is a series of choices that each move a venue up or down. ## A ranking is a choice with a number attached Every published volume figure starts with four decisions: which pairs count, which quote currencies are converted, how long the sample runs, and what gets filtered out as non-economic activity. Change any one of them and the order changes. None of these choices is usually hidden. They are simply rarely read. A reader sees a number, compares it with a number from another ranking, and concludes that one source is lying. More often both sources are internally consistent and externally incomparable. Our own board is not exempt. It ranks 24 venues on sampled 24h volume as of October 2026, and the top row carries roughly a third of that sample. That ordering is a statement about one sample under one method, not a verdict about which venue is largest in every sense. ## Wash trading inflates the numerator Volume is the easiest metric to manufacture, because every trade has two sides and a venue can be on both. Round-tripping between accounts under common control produces turnover with no change in economic exposure. Filtering it is a judgement call. Rules that flag self-matching, identical sizes and sub-second round trips remove some of it. They also remove legitimate market making, which looks similar from outside. Set the filter tight and you delete real flow. Set it loose and you count fake flow. The consequence for the reader is simple. A venue’s volume is an estimate produced by a filter, and the filter is part of the estimate. Two rankings with different filters will disagree about the same venue without either being dishonest. ## Pair universe: spot, perpetual, or both ### Spot only Spot volume counts delivered trades. It excludes leveraged exposure, which is most of the activity on many venues, so a spot-only ranking will rank derivatives-heavy venues lower than their overall business suggests. ### Perpetual only Perpetual volume counts contracts, and the notional depends on how the contract is specified and whether the figure counts one side of each trade or both. A venue that reports one side reports half of what a venue reporting both sides reports for identical activity. ### Both, summed Summing spot and perpetual is common and produces a number that no single order book supports. It is a market-share figure. Read as liquidity, it overstates by the amount of leverage in the mix. ## Quote currency and the sampling window ### Conversion inflates the total Most pairs quote in stablecoins, and stablecoin supply reached $292.40B in October 2026, per RWA.xyz. When a venue reports volume in USD, it converts every quote currency at a rate taken at a time. Two venues can therefore disagree from the conversion step alone. A pair quoted in a local currency trading at a premium converts into a larger USD figure than the same trade quoted in dollars. The premium is a friction price, but the ranking records it as volume. ### The window is a definition too A rolling 24-hour window and a calendar-day window are not the same measurement. The rolling window spreads a venue’s bad hour across the day. The calendar window puts it in one bucket and moves on. Sessions matter as well. A venue whose flow is concentrated in one region will show a different 24-hour figure depending on when the window closes relative to its local trading hours. | Choice | Effect on rank | What the reader can check | |—|—|—| | Pair universe | Derivative-heavy venues rise when perps are included | Whether spot and perp are summed or shown apart | | Wash filter | Loose filters favour venues with round-tripping flow | Whether the filter rules are published | | Quote currency | Local-premium pairs inflate USD totals | Which conversion rate and which date | | Sampling window | Rolling windows smooth spikes that calendar days expose | Whether the window definition is stated | | Double counting | Counting both sides doubles a venue’s figure | Whether one side or both are counted | ## What a disagreement means If two rankings disagree by a position or two, the likely cause is a methodological choice, not fraud. If they disagree by an order of magnitude, look at the pair universe first. Perp inclusion is the largest single source of gap between otherwise similar methodologies. The useful response is to read both methodologies and see which choices differ, rather than to pick the flattering ranking. ## Verdict A ranking is a methodology, and a methodology is a set of checkable choices. Volume figures are estimates produced by filters, pair definitions and conversion rates. They are useful for comparison inside one method and misleading across methods. Read the [methodology](/methodology/) before you quote a rank. Then check the [monthly fee page changes](/fee-pages-change-monthly/), because a venue’s fee tier explains part of why its volume mix looks the way it does. Neither number is a fact about liquidity. Both are facts about a sample. ### FAQ Q: Does a ranking disagreement mean one source is fake? A: Not usually. Most disagreements trace to a documented choice: whether perpetuals are summed with spot, whether both sides of a trade are counted, which conversion rate is applied to non-USD quotes, and how long the sampling window runs. Two honest methodologies will disagree on rank without either being wrong. Q: Why do wash filters make volume uncertain? A: Round-tripping between accounts under common control looks similar to legitimate market making: repeated trades, similar sizes, short holding periods. A tight filter deletes real flow, a loose one counts fake flow, and every published figure sits somewhere between the two. Q: What should I check first when two rankings differ by a lot? A: Check the pair universe before anything else. Whether perpetuals are included, and whether both sides of each contract are counted, moves a figure by more than most other methodological choices combined. ## Staking rewards: income at receipt, gain on sale URL: https://www.caijing188.com/staking-rewards-tax/ Updated: 2026-10-27T22:09:48+08:00 Verified: 2026-10-27 ### Quick answer The value of a staking reward when it lands becomes your cost basis, so the timestamp on the reward record matters more than the annual rate, as of October 2026. ETH staking paid 2.63% gross in Q3 2026, and over 96% of it came from issuance (Bitwise, Q4 2026). This is not tax advice. ### Key takeaways - The value at receipt becomes the cost basis, so the timestamp is the load-bearing field - ETH staking paid 2.63% gross in Q3 2026, within a 2.6% to 3.3% range for the year - Over 96% of ETH rewards and about 90% of SOL rewards came from issuance (Bitwise, Q4 2026) - 43.8M ETH was staked on 23 October 2026, with an exit queue of 850,000 ETH, about 14.8 days ![A staking dashboard showing a reward event with a date and a token amount](https://img1.ladyww.cn/picture/Picture00423.jpg) A staking reward arrives as a quantity of tokens at a moment in time. Two things happen after that. The value at receipt is generally treated as income, and any later change in value is a gain or a loss when you dispose of the tokens. The number you write down at receipt becomes the cost basis for everything that follows, which is why the timestamp matters more than the annual rate. ## Two events, two numbers The first event is receipt. You are given a quantity, and that quantity has a fiat value on that day. The second event is disposal, whenever you sell or swap. The difference between the two values is the capital side. Collapsing the two into one line, usually an annual total in a statement, loses the only input the calculation needs. It is not a reporting inconvenience. It is the disappearance of the number that determines both legs. This is not tax advice, and the treatment differs by jurisdiction. The record-keeping point holds regardless of which treatment applies. ## The timestamp is the argument ### What a lump-sum rewards line hides A venue that reports one rewards total for the month tells you the quantity and, at best, an average price. It does not tell you when each reward landed, so it cannot tell you what each one was worth. If the asset moved sharply inside the month, the average is a price that never existed on any single day. Using it as a cost basis is an estimate presented as a fact. ### What a per-event line gives you A per-event record gives quantity, asset, date, and often a price. That is enough to compute income per event and to carry a cost basis forward. It also makes the disposal leg simpler, because you can match units to lots. Ask for the format before you need it. A CSV export with a date column is worth more than a PDF summary with a total. ## Most of the yield is issuance Over 96% of ETH staking rewards and about 90% of SOL rewards came from new issuance rather than from transaction fees, according to Bitwise’s Q4 2026 review. That single fact changes how a reward should be read. You are not being paid out of revenue a business earned. You are being paid in newly created units whose price is set by the market afterwards. The yield is a policy parameter as much as it is a return. Which means the fiat value at receipt can be higher than the value at sale without anything going wrong at the venue. The income was recorded at the higher number and the disposal realises the lower one. For a position sized in dollars, that is a real cost, and it is invisible in a headline yield. ## Reading the gross number ETH staking paid 2.63% gross in Q3 2026, inside a 2.6% to 3.3% range for the year (Ethereum staking data, Q3 2026). Solana paid 5.99% gross in the same quarter, with validator commission of 5% to 10% leaving 4.6% to 7.6% net for a delegator. Provider fees then take another slice. A 10% protocol fee, a commission near 35% at one large venue, near 39.95% at another, and a 10% to 26% range at a third, are published figures from Q3 2026 fee schedules. Net figures are struck on each provider’s own gross basis and period, so a published net can sit above a quarterly gross print. Compare like with like, and note the period. Solo validation needs 32 ETH, and the queues are part of the cost. The entry queue was about 2.5M ETH on 22 July 2026, more than 43 days. The exit queue was 850,000 ETH in October 2026, about 14.8 days, against 43.8M ETH staked on 23 October 2026. | Record you need | Why it matters | Where it usually lives | |—|—|—| | Reward date | Sets the fiat value at receipt | Per-event export, not a monthly PDF | | Quantity and asset | The unit count you were paid | Rewards tab, distinct from the balance | | Price on that date | The income figure and the cost basis | Venue export, or a daily close from a stated source | | Fee taken | Gross to net, and who took the difference | Fee schedule, not the rewards tab | | Disposal date and price | The second leg of the calculation | Trade history, not the rewards tab | ## Verdict The yield is the least interesting number on the page. 2.63% gross for ETH and 5.99% for SOL are policy outputs, and the share of them that arrives as issuance tells you the price risk sits with you after receipt. What you control is the record. Get a dated, per-event export, keep the fiat value you used, and reconcile it once a quarter rather than once a year. For the yield question itself, [staking yield is not income](/staking-yield-is-not-income/) covers why the gross number overstates what a holder keeps. The [venue board](/exchange/) is the place to check a provider’s reporting before its headline rate, because a venue that will not produce a dated export is a venue you cannot reconcile. ### FAQ Q: Why does the timestamp on a reward matter so much? A: Because the fiat value at the moment of receipt is normally what you record as income, and it is also the number that becomes your cost basis for the later disposal. Without a timestamp you cannot establish either. A daily or per-event record fixes the value; a monthly total leaves you estimating. Q: Does it matter that most staking yield is issuance? A: It matters for how the position behaves. Rewards paid in newly issued units are not a share of revenue the network earned, so their fiat value is set after receipt and can fall before you sell. The income may be recorded at a price you never realise, which is a timing problem rather than an accounting one. Q: Is a higher-fee staking provider always worse? A: On yield alone, usually yes, because the fee comes off the gross rate. But the fee is not the only variable: a provider that supplies clean per-event records, a clear gross figure and a stated period saves you work that has a cost of its own. Compare net, and compare what comes with it. ## Bridges are three products, and only one of them is canonical URL: https://www.caijing188.com/three-kinds-of-bridges/ Updated: 2026-10-27T22:09:49+08:00 Verified: 2026-10-27 ### Quick answer Three products share the word bridge: the rollup's canonical bridge, a third-party liquidity network, and a wrapped-asset issuer. Each holds keys differently and pays you differently. Ethereum.org listed, as of October 2026, 7-day canonical exits on Base and OP Mainnet and a 45,818-block window on Arbitrum. Ask what claim you hold before you move size. ### Key takeaways - Ethereum.org's October 2026 directory puts canonical exits at 7 days on Base and OP Mainnet - Arbitrum's canonical window is 45,818 blocks, about 6.4 days, in the same listing - FeeEdge's 17 August 2026 survey priced L1 ERC-20 withdrawals at $2 to $10, above $50 under congestion - A fast route is fast because someone else is holding the wait, and that someone is a counterparty ![Three bridge types drawn as three separate paths](https://img1.ladyww.cn/picture/Picture00413.jpg) Most bridge losses are not failures of cryptography. They are category errors. A user believes they hold a claim on a rollup and actually hold a claim on a company, or believes a route is official because it is the one their wallet offered first. The three products below look identical in a balance and behave nothing alike when something breaks. ## Product one: the canonical bridge The canonical bridge is part of the rollup’s own contracts. You deposit into a contract on Ethereum, the rollup credits you on its own ledger, and to leave you submit a withdrawal that the rollup’s proof or challenge process has to clear. The keys here belong to the rollup, not to a company. An optimistic rollup releases your withdrawal after its challenge window closes, which is why Ethereum.org’s L2 directory in October 2026 shows Base and OP Mainnet at 7 days to mainnet and Arbitrum’s dispute window at 45,818 blocks, about 6.4 days. A ZK rollup proves the state instead of waiting, and ZKsync Era quotes a 3-hour minimum in the same October 2026 Ethereum.org listing. What you receive is a native claim. If the rollup is live and its contracts have not been upgraded into something else, that claim is enforceable on mainnet without anyone’s permission. ## Product two: the liquidity network A third-party bridge is a market maker with inventory on both sides. You send funds to its contract on one chain, and it pays you from its pool on the other, usually within minutes. The keys are the network’s contracts and its treasury. Nothing here waits for the rollup, which is precisely the product: the pool absorbs the wait and prices it into a spread or a fee. You receive the native asset, but it arrives because a business decided to send it, not because a proof cleared. This is not automatically worse. It is faster, it works across chains that share no settlement layer, and for small amounts the spread can undercut a canonical exit. What it is not is canonical. FeeEdge’s survey of 17 August 2026 priced these routes differently by chain, with Solana under $0.01, Base at under $0.01 to $0.10, Arbitrum at $0.01 to $0.50, Tron at $1 to $3, and Bitcoin at $0.50 to $5 and above. The spread across chains is a liquidity fact, not a safety signal. ## Product three: the wrapped-asset issuer The third product issues a token that represents a claim on an asset held somewhere else. You hold a symbol, and behind it sit a custodian, a redemption desk, and a legal agreement. The keys are the issuer’s mint authority and its custody arrangement. What you receive is an IOU. It trades like the underlying, until you want the underlying, at which point the redemption process applies. That process involves identity checks at most issuers, a queue, and often a minimum size. The failure mode is not a smart contract bug. It is the issuer stopping, the custody going wrong, or the redemption terms changing while you hold the token. | Product | Who holds the keys | What you actually hold | Clock to exit | |—|—|—|—| | Canonical rollup bridge | The rollup’s contracts and its proof or challenge rule | A native claim enforceable on mainnet | 7 days on Base and OP Mainnet, about 6.4 days on Arbitrum, 3 hours minimum on ZKsync Era, per Ethereum.org in October 2026 | | Third-party liquidity network | The network’s contracts and its inventory | A transfer paid from a pool | Minutes, because the pool carries the wait | | Wrapped-asset issuer | The issuer’s mint authority and custodian | An IOU redeemable through the issuer | Redemption queue plus identity checks at most issuers | | Exchange withdrawal | The venue | A credit on the venue’s ledger | Venue policy, not a chain rule | ## Four questions before you move size What claim do I hold? Native, pooled, or issued. If the answer is not native, you have a counterparty. Know its name. Who can stop me? Find the key that can pause, freeze, upgrade or blacklist. A bridge whose contracts can be upgraded by a multisig is a bridge with an operator, whatever the documentation says. What is the exit clock, and who pays for it? A 7-day canonical exit, the Ethereum.org figure for Base and OP Mainnet in October 2026, is free of counterparty risk and expensive in time. A minutes-long pooled exit is the reverse. Both are prices, and only one of them appears in the fee. What happens if the operator stops? For a native claim, nothing. For a pool, your transfer stops. For an issuer, redemption becomes a legal process with an unknown timeline. ## Verdict Only the canonical bridge gives you a claim on the rollup itself. The other two give you a claim on a business, and businesses are useful, fast, and occasionally gone. Size accordingly. Move small amounts through the fast routes and pay the spread. Move size through the route whose failure mode you have read, and check the exit clock first. Start with [what a sequencer does](/what-a-sequencer-does/) to see who produces the state your claim depends on, then check the [DeFi section](/category/defi/) for how those claims get reused once they land. ### FAQ Q: Is the bridge my wallet suggests the official one? A: Wallets rank routes by convenience and by commercial arrangement, not by claim quality. The canonical bridge is the rollup's own contract, while a liquidity network is a separate business that fronts the transfer. Both produce a balance that looks identical, and only one is enforceable against the rollup without a counterparty. Q: What do I actually receive from a third-party bridge? A: A transfer on the destination chain paid out of the network's own pool. The pool holds inventory on both sides, so it can pay you before the canonical message settles. You receive the native asset, but your safety depends on the pool being solvent and on its contracts behaving, which is a different question from whether the rollup is valid. Q: How long should a canonical withdrawal take? A: Long enough to clear the rollup's own settlement rule. Ethereum.org's October 2026 listing shows 7 days for Base and OP Mainnet, 45,818 blocks, about 6.4 days, for Arbitrum, and a 3-hour minimum for ZKsync Era. A route quoting minutes is not the canonical bridge, whatever its interface says. ## Multisig for a small team: the setup that does not lock you out URL: https://www.caijing188.com/multisig-small-team/ Updated: 2026-10-27T22:09:49+08:00 Verified: 2026-10-27 ### Quick answer A 2-of-3 multisig survives one lost key and one compromise, which is why it fits a small team, while 3-of-5 suits a treasury and needs five maintained signers. Staked ETH showed the availability side as of October 2026, with an exit queue of 850,000 ETH, about 14.8 days. Test a recovery before funding. ### Key takeaways - Every extra signature buys theft resistance and spends availability - Three keys in one flat on three devices from one vendor is a 1-of-1 in practice - Most real losses come from procedure: a trusted initiator and an unchecked destination - Staked ETH's exit queue held 850,000 ETH, about 14.8 days, in October 2026, which is the kind of clock a lockout compounds ![Three signing devices on a desk, one kept offsite](https://img1.ladyww.cn/picture/Picture00453.jpg) A multisig replaces one key with a threshold, so no single device can move funds. That is the theft half of the problem. The other half is availability: the same threshold that stops a thief stops you when a signer is lost, a colleague leaves, or a device sits in a country you no longer visit. For a small team the design question is not how many signatures, but how many failures of each kind the wallet can survive. ## The threshold is a trade, not a setting An m-of-n wallet needs m signatures from n keys. Raising m hardens the wallet against a single compromised signer. It also raises the chance that ordinary life locks you out. The arithmetic is simple and usually ignored. A 2-of-3 wallet survives one lost key and one compromised key. A 3-of-5 survives two of either, but now five devices or five people must be maintained, and every payment needs three of them awake. A 4-of-7 looks safer on paper and behaves worse in practice, because coordination cost grows faster than security does. For a small team, 2-of-3 is the default to start from. A treasury with a formal payment process and more than one person able to authorise a transfer belongs at 3-of-5. ## Where the signers live matters more than the count Three keys held by three people in the same room, on devices bought in the same order from the same vendor, is a 1-of-1 with extra steps. Fire, burglary, a legal seizure of one office, or a single firmware flaw takes all three at once. Diversity is the control. Different people, different cities, different device vendors, different firmware versions, and at least one signer that is not a browser extension. A software signer on a laptop that also browses is a hot key with a multisig label on it. Write this down as a matrix: signer, person, device, location, backup, and who can reach it on a Sunday. The matrix usually reveals that two signers share a flat. ## The failure that actually happens is process Most multisig losses in practice come from procedure, not cryptography. The initiator builds a transaction, the other signers approve it because the initiator is trusted, and the destination was wrong or the payload was not what the label said. Four habits close most of it. Read the destination and amount on the signing device, not in the coordination app. Confirm the first and last characters of the address out of band, by voice, for anything above a set limit. Require the initiator to state the purpose in the transaction memo. And name a deputy for every signer, so a holiday does not freeze the wallet. ## Test a recovery before funding Build the wallet, then break it on purpose. Remove one signer, restore it from its backup, and confirm the wallet still spends. Better, run a full drill: rebuild the wallet from the backups alone on fresh hardware, with the original devices switched off. Do this while the balance is small. A recovery you have never performed is a plan, and plans fail at the moment they are needed. Availability has a clock you do not control. Staked ETH is the clearest example: the exit queue held 850,000 ETH in October 2026, about 14.8 days, per this board’s data pack, and a solo validator needs 32 ETH to exist at all. A treasury asset with a redemption window behaves the same way. If a multisig lockout lands inside that window, the delay compounds. | Configuration | Key losses tolerated | Compromises tolerated | Typical use | |—|—|—|—| | 1-of-1 hardware | None | None | Personal savings | | 2-of-2 | None | None | Partnership with no slack | | 2-of-3 | One | One | Small team default | | 3-of-5 | Two | Two | Treasury with a payment process | | 4-of-7 | Three | Three | Larger organisation with dedicated ops | ## What to do next Choose 2-of-3 unless you can name the person who will run a 3-of-5 process every week. Then place the signers so no single event removes two of them, and record the placements somewhere the team can find them without the keys. Run the recovery drill before the first real deposit, and repeat it when a signer leaves. Staff turnover is the most common way a working multisig becomes an unusable one. Finally, decide what each wallet is for. A multisig holding trading inventory and a multisig holding reserves should not share signers, because the first needs speed and the second needs deliberation. The [hot wallet sizing question](/how-much-in-a-hot-wallet/) sets that boundary, and [how a custodian proves reserves](/proof-of-reserves-three-numbers/) covers the part a multisig cannot answer on its own. ### FAQ Q: Is 2-of-3 enough for a company treasury? A: It depends on the process behind it. 2-of-3 tolerates one lost key and one compromised key, which is adequate when three people genuinely control three devices in different places. If one person holds two signers, the effective threshold is 1 and the count on paper is misleading. Q: What happens if a signer leaves the company? A: The wallet keeps working until that signer's key is needed, and then it may not. Replace the signer before the departure, not after, and run a recovery drill with the new set. Turnover is the most common route from a working multisig to an unusable one. Q: Should I use different wallet vendors for each signer? A: Different vendors reduce the chance that one firmware flaw removes several signers at once. Different people, cities and devices matter at least as much. The test is whether a single event, a fire, a seizure or a bad update, can take two signers together. ## What it costs to move $20 on eight networks URL: https://www.caijing188.com/cost-of-twenty-dollars/ Updated: 2026-10-27T22:09:50+08:00 Verified: 2026-10-27 ### Quick answer Moving $20 cost under a cent on Solana and $2 to $10 on Ethereum L1 ERC-20 in FeeEdge's 17 August 2026 survey, or 0.01% to 50% of the amount, as of October 2026. Tron's $1 to $3 works out at 5% to 15% of the same transfer. Batch small moves. ### Key takeaways - Solana under $0.01 and Polygon PoS $0.002 to $0.01 per transfer, on 17 August 2026 - Ethereum L1 ERC-20 at $2 to $10, which is 10% to 50% of a $20 transfer - Tron's $1 to $3 is 5% to 15% of $20, more than five years of 2.63% staking yield at the high end - Exchange withdrawal fees add 0.3 to 1 USDT, or 1.5% to 5% of a $20 withdrawal ![A small parcel weighed against a stack of fee receipts](https://img1.ladyww.cn/picture/Picture00444.jpg) Moving $20 across eight networks cost between under a cent and ten dollars in FeeEdge’s survey of 17 August 2026. As a share of the amount moved, that is 0.01% to 50%. At the top of the range the fee is not a cost attached to the transfer. It is the transfer. ## A flat fee against a small ticket Network fees price computation and block space, not value. A $20 transfer and a $20,000 transfer consume roughly the same resources, so they cost roughly the same. The absolute fee is flat and the percentage is not. That makes the percentage the only honest way to read a fee on a small ticket. Halve the amount and the percentage doubles. The same $5 fee is 25% of $20, 0.5% of $1,000, and 0.05% of $10,000. Most fee comparisons quote the absolute number, because absolute numbers are easy to publish and they flatter the expensive chains. A fee table in dollars answers a question nobody with a small balance is asking. ## Eight networks, one twenty-dollar bill The range in the survey is wide enough to change behaviour. Solana came in under $0.01, which is under 0.05% of $20. Polygon PoS was $0.002 to $0.01. BNB Chain was $0.01 to $0.20. Base was under $0.01 to $0.10. Arbitrum was $0.01 to $0.50. Then the range breaks. Tron was $1 to $3, or 5% to 15% of the transfer. Bitcoin was $0.50 to $5 or more. Ethereum L1 ERC-20 was $2 to $10, with congestion taking it above $50, which is 10% to 50% of a $20 move and more than 250% in the worst case. A $50 fee on a $20 transfer costs two and a half times the amount being moved. The transaction still validates and still confirms. It is simply irrational, and no yield on the board repairs it. ## The recovery arithmetic A fee is a loss you have to earn back before the position is whole. The required gain is not the fee percentage. It is the fee divided by what remains. Pay 10% and you hold $18 of a $20 position, so you need 11.1% to get back to $20. Pay 15% and you need 17.6%. Pay 25% and you need 33.3%. Pay 50% and you need 100%. Set that against the yields on the board. ETH staking paid 2.63% gross in Q3 2026. Tokenised T-bills paid 3.60% to 3.61% annualised on 7 October 2026. A 10% transfer fee is nearly four years of gross staking yield. The 15% that Tron’s high end produces on a $20 move is more than five years of it. That is the comparison that makes a fee decision, and it has nothing to do with which chain is faster. ## The withdrawal fee is the bigger one The network fee is not the only charge on a small transfer. The venue charges too, and on small tickets that charge dominates. FeeEdge’s 17 August 2026 figures put exchange withdrawal fees at about 0.5 to 1 USDT for Solana SPL, roughly 0.3 USDT for BEP-20, and highest for ERC-20. On a $20 withdrawal, 0.5 to 1 USDT is 2.5% to 5% before the network fee is counted at all. Entry-tier spot taker fees sat near 0.10% in the same survey, which is about $0.02 on a $20 trade. Trading is the cheap part. Leaving is the expensive part, and the published fee page usually lists both without ranking them. | Network | Typical fee, 17 August 2026 | Share of a $20 transfer | Share of a $2,000 transfer | |—|—|—|—| | Solana | Under $0.01 | Under 0.05% | Under 0.0005% | | Polygon PoS | $0.002 to $0.01 | 0.01% to 0.05% | 0.0001% to 0.0005% | | BNB Chain | $0.01 to $0.20 | 0.05% to 1.0% | 0.0005% to 0.01% | | Base | Under $0.01 to $0.10 | Under 0.05% to 0.5% | Under 0.0005% to 0.005% | | Arbitrum | $0.01 to $0.50 | 0.05% to 2.5% | 0.0005% to 0.025% | | Tron | $1 to $3 | 5% to 15% | 0.05% to 0.15% | | Bitcoin | $0.50 to $5+ | 2.5% to 25%+ | 0.025% to 0.25%+ | | Ethereum L1 ERC-20 | $2 to $10 | 10% to 50% | 0.1% to 0.5% | ## What to do next Batch. The fee is per transfer, so consolidating five $20 moves into one $100 move cuts the fee bill by four fifths and cuts the percentage by the same factor. Choose the rail by ticket size rather than by habit. Under a few hundred dollars, Solana and Polygon PoS hold the fee below 0.05% of the amount, and Ethereum L1 holds it at 10% or worse. Above a few thousand, the L1 absolute cost stops mattering and its depth starts to. Then check the page before sending. Fees are published per venue and they change, and the numbers above are one survey on one date. Our [fee page tracker](/fee-pages-change-monthly/) exists because a fee quoted in a guide is a fee that has already moved. Compare venues on the [exchange board](/exchange/) using the withdrawal column rather than the trading column, since the withdrawal column is what a small balance actually pays. ### FAQ Q: Why is a $20 transfer so expensive relative to its size? A: Fees price computation and block space rather than value, so a $20 transfer and a $20,000 transfer consume similar resources and cost similar amounts. The absolute fee stays flat while the percentage scales inversely with size. On Ethereum L1 ERC-20, FeeEdge's 17 August 2026 survey put the fee at $2 to $10, which is 10% to 50% of a $20 move. Q: Which network is cheapest for small transfers? A: Solana and Polygon PoS lead in the 17 August 2026 survey, at under $0.01 and $0.002 to $0.01 respectively, which is 0.05% or less of a $20 transfer. Base at under $0.01 to $0.10 and BNB Chain at $0.01 to $0.20 come next. Tron, Bitcoin, and Ethereum L1 ERC-20 all exceed 2.5% of the amount. Q: How much does an exchange withdrawal fee add? A: FeeEdge's 17 August 2026 figures put withdrawals at about 0.5 to 1 USDT for Solana SPL and roughly 0.3 USDT for BEP-20, with ERC-20 highest. On a $20 withdrawal, 0.5 to 1 USDT is 2.5% to 5% before the network fee. Entry-tier spot taker fees near 0.10% are about $0.02 on the same $20, so the exit costs far more than the trade. ## Reading a vault APY: base yield, incentives, and the part that expires URL: https://www.caijing188.com/vault-apy-anatomy/ Updated: 2026-10-27T22:09:50+08:00 Verified: 2026-10-27 ### Quick answer A vault APY usually adds three legs: base yield, an incentive token, and points. Only the base is a rate, and only the net after fees and slippage compounds. Maple paid 5.20% and Plume nOPAL 10.73% against 3.60% to 3.61% on tokenised Treasuries, as of October 2026, which is credit risk rather than a risk-free rate. ### Key takeaways - Maple quoted 5.20% and Plume nOPAL 10.73% in RWA.xyz readings for October 2026 - Tokenised Treasuries quoted 3.60% to 3.61% annualised on a seven-day basis on 7 October 2026 - 10.73% is more than 7 points above a Treasury, which is the price of credit risk - Private credit was about $8 billion of the on-chain RWA book in October 2026 ![A vault page splitting a headline APY into base yield, incentives and points](https://img1.ladyww.cn/picture/Picture00434.jpg) A vault APY is a sum of things that behave differently. One part is cash flow from borrowers, one part is a token somebody is paying you with, one part is a promise with no price. Adding the three into one number and comparing that number to a Treasury bill is the most common mistake in the category. ## The headline is a sum, not a rate Pull apart any vault page quoting a double-digit APY and three legs usually fall out. A base yield paid in the deposited asset, an incentive paid in a governance token, and a points programme that pays in nothing yet. Only the first is a rate. The second is a transfer of a volatile asset, sized at today’s price. The third is an option. The arithmetic is addition. The economics is not, because the legs differ in duration, failure mode, and payer. ## The incentive leg is paid in the asset you are underwriting Take a vault quoting a base yield plus a token reward. The reward is quoted at the token’s current price and paid over a fixed programme window. Two things change the moment you deposit. The reward is denominated in a token you now hold, so you have taken a long position in the issuer’s governance asset on top of your exposure to the vault. And the quoted value of that reward falls if the token falls, while the number of tokens you receive does not rise to compensate. Selling as it arrives is the obvious answer and it adds a second cost. The token is usually thin relative to the programme’s size, and your clip is one in a stream of sellers on the same schedule. The realised value of an incentive sits below the quoted value by the slippage of the whole programme. ## Points are not a rate Points are the least yield-like leg. A point balance is an unpriced claim on a token that may or may not exist, distributed by a formula that can change, at a valuation nobody has set. None of that is necessarily dishonest. It is simply not a rate, and a column beside rates makes it comparable to things that are. The right treatment is an option: some probability of some value at some date. There is a structural point about who pays. An incentive programme is a customer acquisition cost settled in equity, and budgets end. The base yield does not, because it comes from a borrower who is still paying. ## Realised yield is the only number that compounds The number that compounds is what lands in your wallet, in the asset you intend to keep, after every cost between the vault’s gross return and your balance. That means subtracting management and performance fees, the cost of entry and exit, slippage on converting a non-native reward, and the gas or bridging to move in and out. | Component | Paid in | Expires | Counts toward realised yield | |—|—|—|—| | Base yield | The deposited asset | No, unless the borrower leaves | Yes, at face value | | Incentive tokens | The issuer’s token | Yes, when the programme ends | Only after sale, net of slippage | | Points | Nothing yet | Yes, and unilaterally | No, until something prices them | | Fees and slippage | Subtracted | No | Negative, always | The same discipline applies to the fee and rate pages themselves, which is why we read them as trailing evidence rather than forward guidance ([fee pages change monthly](/fee-pages-change-monthly/)). ## Credit risk is not a spread Maple’s 5.20% and Plume’s nOPAL at 10.73%, both in RWA.xyz’s October 2026 readings, are not risk-free rates. They are credit spreads presented as yields. The reference point is the tokenised Treasury complex at 3.60% to 3.61% annualised on a seven-day basis on 7 October 2026. Against that, 5.20% is roughly 1.6 points of compensation for lending to private borrowers, and 10.73% is more than 7 points. Private credit was about $8 billion of the on-chain RWA book in October 2026, and it is a market where a borrower can default. A 10.73% headline on a private credit vault is not a high-yield savings account. It is the mean of a distribution that includes a loss, and the loss does not appear in the APY column. Yield paid in issuance deserves the same scepticism ([staking yield is not income](/staking-yield-is-not-income/)). ## What to do next Rebuild the number before comparing it. Treat the base yield as the only rate, discount the incentive for the slippage of the whole programme, and carry points at zero until something prices them. ## Verdict A vault APY is a marketing number assembled from three legs with different durations. The base yield is a rate, the incentive is a position, and the points are an option. Only the first compounds on its own, and only the net of all three is yours. Compare credit spreads with credit spreads. Maple at 5.20% and nOPAL at 10.73% belong beside a corporate bond ladder, not beside a T-bill. The gap to a tokenised Treasury at 3.60% to 3.61% is the price of that distinction. ### FAQ Q: Why is a 10.73% vault not comparable to a 3.6% Treasury? A: Because one is a credit spread and the other is close to a risk-free rate. The extra yield compensates for borrowers who can default, capital that can be locked, and an exit that may take longer than a Treasury settlement. Comparing them directly overstates the vault by most of the spread. Q: Are points worth anything in an APY calculation? A: Not until something prices them. Points are an unpriced claim on a token that may not exist, distributed by a formula that can change. Treating them as zero is not pessimism, it is the only treatment that keeps the headline comparable to a rate. Q: What costs sit between the headline and the realised yield? A: Management and performance fees, entry and exit costs, slippage when converting a non-native reward, and the gas or bridging needed to move in and out. On a large position these are small. On a small one they can consume the entire incentive leg. ## The same coin costs more in some markets, and it is not arbitrage URL: https://www.caijing188.com/regional-premium-pricing/ Updated: 2026-10-27T22:09:51+08:00 Verified: 2026-10-27 ### Quick answer A local premium is a friction price, set by banking rails, capital controls and withdrawal windows rather than by the coin. Stablecoin supply reached $292.40B as of October 2026 per RWA.xyz. Capturing the gap needs a local bank account, so most retail traders outside the market cannot. ### Key takeaways - A premium is a friction price paid by whoever needs the local rail - Stablecoin supply of $292.40B in October 2026 per RWA.xyz is gross across chains, not local float - Base and OP Mainnet exits take seven days per Ethereum.org's 1 to 6 October 2026 snapshot, long enough for a gap to close - Fixed costs mean a small premium on a small ticket often closes at a loss ![The same coin priced differently on three local exchange screens](https://img1.ladyww.cn/picture/Picture00404.jpg) The same coin trades at different prices in different markets, and the gap can persist for months. That looks like free money and mostly is not. The premium is the price of a friction: a local banking rail, a capital control, or a withdrawal window that closes exactly when you need it. Stablecoin supply reached $292.40B in October 2026, per RWA.xyz, and where that float sits decides whether a premium is reachable. ## The premium is the price of a rail A price gap is only arbitrageable if you can move the asset and move the money. On a local venue, the second leg is a bank account in the local currency, opened under local identity rules. That is the part a foreign trader does not have. The premium exists because the local rail is scarce. Deposits clear slowly, withdrawals are capped, or the correspondent bank applies scrutiny that makes a large transfer expensive. Each of those frictions lifts the local price and holds it there. A premium is usually quoted as a percentage of the global price, and it reads better as a spread than as a level. It widens when local demand rises and narrows when local supply arrives, so a gap that has persisted for months is telling you that supply has not arrived. That is a statement about the rail, not about the coin. ## Why the trade does not close The textbook version of the trade is to buy where it is cheap, sell where it is expensive, and repeat until the gap closes. The practical version meets a queue at every step. Moving the coin is the easier leg and it is not free. FeeEdge’s survey of 17 August 2026 put exchange withdrawal fees at roughly 0.5 to 1 USDT for Solana SPL, about 0.3 USDT for BEP-20, and highest for ERC-20, with L1 ERC-20 transfers at $2 to $10 and above $50 in congestion. Moving the fiat is the hard leg. It needs a local account, local tax treatment, and a counterparty willing to take the other side at a size that matters. A retail trader typically has none of the three. Size is the other constraint. Fixed costs do not scale with the ticket, so a premium has to be large relative to the withdrawal fee, the funding cost and the spread before a round trip clears anything. On a small balance the arithmetic often closes at a loss even when the gap is real. ## What stress does to the gap Premiums widen under stress, and that is when they are least capturable. When a venue suspends withdrawals, the local price decouples from the global one, because the asset cannot leave. The decoupling is not a signal about the coin. It is a signal about the exit. Anyone who buys the local discount or sells the local premium during a suspension is taking a position on when the exit reopens, which is a different trade from the one they think they are making. Bridging is a partial substitute with its own clock. The Ethereum.org L2 directory, in its 1 to 6 October 2026 snapshot, lists a seven-day mainnet withdrawal for Base and OP Mainnet, a dispute window of 45,818 blocks, roughly 6.4 days, for Arbitrum, and a minimum of three hours for ZKsync Era. A seven-day delay is long enough for a premium to close before the funds land. ## Float location is the real variable $292.40B of stablecoin supply, per RWA.xyz in October 2026, is a gross figure across every chain. The part that can settle in a given local market at a given hour is much smaller, and that fraction is what sets the local price. A market with deep local float carries a small premium. A market that has to import float carries a larger one, because every import pays the bridge fee, the venue withdrawal fee and the delay. | What the premium needs | What it costs | Who can pay it | |—|—|—| | Local bank account in the quote currency | Identity and residency requirements | Local residents and entities | | Deposit and withdrawal rails that clear at size | Caps and scrutiny on large transfers | Established local accounts | | Float on the local chain | Bridge and withdrawal fees, plus delay | Anyone, at a price | | A counterparty at size | Spread and market impact | Desks with local credit | | Time for the gap to persist | Funding cost of the position | Patient capital only | ## Verdict The premium is a friction price, and frictions are paid by whoever needs the rail. A retail trader outside the local market usually cannot capture it, because the binding constraint is the bank account rather than the coin. Treat a persistent premium as evidence about a local rail, not as a mispricing in the coin. Check where the float is before assuming a gap is tradeable: the [stablecoin supply breakdown](/stablecoin-supply-who-holds-it/) explains why a global supply figure says little about local depth. For venue-level exit costs, start with the [venue directory](/exchange/). ### FAQ Q: Is a regional premium always arbitrage? A: No. Arbitrage requires moving both the asset and the money. The asset leg is cheap and fast on most chains. The money leg needs a local bank account, local tax treatment and a counterparty at size, which is why the gap persists rather than closing. Q: Why do premiums widen when a venue suspends withdrawals? A: A suspension breaks the link between the local price and the global one, because the asset cannot leave. The resulting gap is a price on the exit reopening, not a new view on the coin. Sellers of the local premium are taking a timing position they may not have intended. Q: What does stablecoin float location change? A: It sets how much of a market's supply can settle locally at a given hour. RWA.xyz put total supply at $292.40B in October 2026, but the share held on a given chain or in a given banking system is what determines local depth and therefore the local premium. ## KYC tiers change your withdrawal limit more than your privacy URL: https://www.caijing188.com/kyc-tiers-withdrawal-limits/ Updated: 2026-10-27T22:09:51+08:00 Verified: 2026-10-27 ### Quick answer Verification tiers gate withdrawal limits and payment rails, not just identity, as of October 2026. Our board tracks 24 venues, and a Solana SPL withdrawal costs about 0.5 to 1 USDT while an ERC-20 withdrawal costs $2 to $10 in FeeEdge's 17 August 2026 survey. Read the tier table before you fund. ### Key takeaways - A withdrawal tier changes your rails and your daily cap, not only the documents you upload - Solana SPL withdrawals cost about 0.5 to 1 USDT, ERC-20 withdrawals $2 to $10 (FeeEdge, 17 August 2026) - A no-KYC venue can match a trade but usually cannot pay out in fiat - Our board ranks 24 venues as of October 2026, and tier tables are not comparable across them ![A tiered verification table beside a withdrawal screen](https://img1.ladyww.cn/picture/Picture00424.jpg) Verification is usually sold as a privacy trade-off. In practice it is a payments decision. The tier you are in decides which withdrawal rails you can use, how much can leave in a day, and whether fiat can leave at all. The identity check is the entry fee. The tier table is the product. ## Verification is a payment decision A venue can rarely offer a bank payout to a customer it has not identified, because the bank or payment provider on the other side has its own obligation to know the beneficiary. So the identity check is not primarily a database entry. It is the thing that makes a fiat rail possible. That is why the same exchange can feel generous at the top of the table and restrictive at the bottom. The difference is not caution. It is which intermediaries will accept the instruction. ## The tier table is the real fee schedule ### Where the limits sit Tier tables usually publish a daily withdrawal cap, a monthly cap and a list of supported networks per tier. The caps are commonly quoted at the highest tier, which is the number the marketing page repeats and the number most readers will never reach. Read the row that matches the documents you are willing to provide. That row, not the headline, is your account. ### What the top tier unlocks The top tier typically adds fiat rails, larger caps and sometimes an account manager for large transfers. It may also add a review step: a large withdrawal can be held for source-of-funds checks that the lower tiers never trigger. More access is not the same as less friction. Ask what the review looks like before you need it. A cap you can hit but cannot clear is worse than a lower cap you understand. ## Why a no-KYC venue usually cannot pay fiat An exchange that accepts no identity documents can still match a trade. It cannot easily hold a bank account that pays out to an unnamed beneficiary. So the restriction shows up as an absence: no fiat rail, no card, no local payment method. In that model your exit is a transfer to another venue or to a self-custody address. That is a real exit, and it costs what the network costs. A Solana SPL withdrawal runs about 0.5 to 1 USDT and an ERC-20 withdrawal $2 to $10 in FeeEdge’s 17 August 2026 survey, above $50 in congestion. Those are the tolls on the only door you have. The second-order cost is the receiving venue. A deposit arriving from an unverified counterparty can trigger its own review, which turns a cheap exit into a slow one. ## Exit cost moves with the tier Higher tiers usually unlock cheaper rails, because a verified account can use a batched internal transfer or a local scheme. That is a fee effect, not a privacy effect. The network choice matters as much. The same dollar can leave for cents on Solana, for about 0.3 USDT on BEP-20 or for $2 to $10 as an L1 ERC-20 transfer, all in FeeEdge’s 17 August 2026 survey. A tier that only exposes the expensive network is a tier with a hidden cost. Our board ranks 24 venues as of October 2026 and the [fee pages change monthly](/fee-pages-change-monthly/), so the number you check today is a snapshot. Confirm the network before you confirm the amount, and confirm which tier the quoted fee applies to. | Gate | What it typically unlocks | What it does not change | |—|—|—| | Email only | Read access, small crypto withdrawals | No fiat rail, no card payout | | Identity document | Fiat deposit and withdrawal, higher daily cap | Network fees on the crypto rail | | Proof of address | Local payment schemes, higher monthly cap | Review time on a large transfer | | Source of funds | Large withdrawals cleared for release | The venue’s obligation to ask again | | Full institutional | OTC settlement, named custody arrangements | Counterparty risk on the venue itself | ## What to do next Open the tier table and find the row you are actually in. Note the cap, the networks and whether fiat appears. Then check the [venue board](/exchange/) for the rail you intend to use, because a tier that supports a network is not the same as a venue that supports the destination. If the answer is crypto-only, size the exit as a network cost rather than a bank cost. That is the honest comparison, and it changes which venue looks cheap. Privacy is the smaller part of this. The larger part is that verification decides where your money can go, and the table that says so is published, usually one click from the fee page. ### FAQ Q: Does a higher KYC tier reduce my fees? A: It can, indirectly. Higher tiers usually unlock cheaper rails, such as batched internal transfers or local payment schemes, and larger limits that make a single withdrawal cheaper than several small ones. The network fee itself does not change with your tier, so the saving comes from the rail and the batching, not from the verification. Q: Why can a no-KYC exchange not pay me in fiat? A: Because the payout runs through a bank or payment provider with its own obligations, including knowing who the beneficiary is. A venue that cannot identify its customer struggles to satisfy that. The restriction appears as an absence: no bank transfer, no card payout, no local scheme. Crypto withdrawal still works. Q: What should I check in a tier table first? A: The daily and monthly withdrawal caps, the list of supported networks, and whether fiat withdrawal appears at any tier. Then check what each tier requires, because a limit you cannot reach without documents you do not have is not a limit you hold. The headline number is usually the top tier, not yours. ## MEV on rollups: the sequencer sees your transaction first URL: https://www.caijing188.com/mev-on-rollups/ Updated: 2026-10-27T22:09:52+08:00 Verified: 2026-10-27 ### Quick answer A single sequencer holds the ordering right, so a rollup swap is not a public mempool race but a private queue with an operator's policy attached. Ethereum.org's snapshot, as of October 2026, listed Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction. Ask what mechanism backs a protection claim before trusting the label. ### Key takeaways - A single sequencer sees the transaction before it is public, so ordering is granted rather than auctioned - Ethereum.org's 1 to 6 October 2026 snapshot put Base at $0.001 to $0.002 per transaction and Arbitrum One at $0.004 to $0.005 - FeeEdge's 17 August 2026 survey put entry-tier spot taker fees near 0.10%, far above an L2 execution fee - MEV protection is a mechanism claim, so ask for the mechanism and not the slogan ![A single sequencer ordering a queue of transactions](https://img1.ladyww.cn/picture/Picture00414.jpg) On a rollup, ordering is decided by one party, and that party reads your transaction before it becomes public. The familiar picture of MEV, with searchers racing in an open mempool, describes a queue most rollup users never join. What replaces it is a private queue with an operator’s policy attached, and the policy is not enforced by the protocol. ## The race moved indoors On Ethereum, a transaction is broadcast to a mempool anyone can read. Searchers watch it, work out whether reordering it pays, and bid for position. The competition is ugly, but it is a competition, and priority has a visible price. A rollup with one sequencer removes that arena. Your transaction goes to the operator, the operator decides where it lands in the next block, and no third party gets to bid for the slot ahead of you. In one reading that is protection. In another it is a monopoly on ordering handed to a single firm, with terms set in a policy document rather than in code. Both readings are correct at once. What changed is not how much ordering is worth, but who gets to compete for it. ## What a swap actually looks like You submit a swap. Before inclusion the operator knows the pair, the size, the direction and the minimum output you set. On a chain where anyone can read the queue, that is exactly the information a sandwich needs. The difference is what happens next. In a public mempool, another party can pay to get in front of you. In a private queue, only the operator can, and whether it does depends on its policy, its reputation and its revenue model. So a rollup swap is not a bidding war against bots. It is trust in an operator’s ordering policy, extended for the few seconds between submission and inclusion. For a small trade that trust is free. For a large one it is the whole cost. Execution fees no longer discipline this. Ethereum.org’s L2 directory, in its 1 to 6 October 2026 snapshot, listed Base at $0.001 to $0.002 per transaction and Arbitrum One at $0.004 to $0.005. At that price, priority bidding on gas is not where anyone competes. Ordering is the scarce resource, and ordering is not sold in the fee market. ## Protection needs a mechanism The phrase MEV protection appears on most rollup landing pages. It is a claim about a mechanism, so it can be checked by asking which one. Commit-reveal is one. You submit a hash of the transaction, the operator orders the hash without knowing its contents, and you reveal afterwards. It removes the operator’s informational edge at the cost of an extra round trip. Encrypted submission is another. The transaction stays sealed until it is ordered, then decrypts in the block. It solves the same problem with different latency and a different key holder. A published ordering rule is a third. First in, first out is a mechanism if it is applied to everything and if someone can audit the log. It is a slogan if it is applied when convenient. More than one party able to order is the fourth, and the only one that removes the single point rather than constraining it. A chain with several sequencers, or with a forced-inclusion path back to mainnet, cannot quietly hold your transaction. | Claim on the page | Mechanism that would make it true | What to check before trusting it | |—|—|—| | No frontrunning | Commit-reveal or encrypted submission | Is there a gap between submission and visibility, and who holds the key | | Fair ordering | A rule applied to every transaction | Is the rule published, and is the ordering log auditable | | MEV rebates | An escrow that returns captured value | Who measures the capture, and against what benchmark | | Decentralised sequencing | More than one party can order | How many can order, and can any single one censor | | No sandwiches | A minimum-output check plus ordering control | Whether the check lives in the protocol or in the interface | ## Why concentration matters more here The bigger the chain, the more ordering rights are worth. Ethereum.org’s October 2026 directory put Base at $14.6B in value and Arbitrum One at $12.3B, with OP Mainnet at $1.65B. A single operator deciding the order of transactions on a chain of that size is a concentrated position, and concentration is what turns a policy question into a governance question. It also explains why the debate is about who runs the sequencer rather than how much MEV exists. The number is hard to observe from outside. The operator list is not. ## Verdict On a rollup you are not outbidding bots. You are trusting an operator’s policy, briefly and completely. That is an acceptable trade for retail size on a chain with a credible operator and a forced-inclusion path. It is a poor trade for size you cannot afford to have reordered, on a chain whose ordering rule you cannot read. Read [what a sequencer does](/what-a-sequencer-does/) for the mechanics, and [who secures a rollup](/who-secures-a-rollup/) for who could replace the operator if it stopped behaving. Until a mechanism is named, treat MEV protection as a label rather than a feature. ### FAQ Q: Is there less MEV on a rollup than on Ethereum? A: The competition is lower and the concentration is higher. On Ethereum, searchers bid in a public mempool and the winner pays for priority. On a single-sequencer rollup, one operator sees the queue and decides the order, so extraction depends on its policy rather than on an auction, as of October 2026. Q: Why does gas price not protect me on an L2? A: Because execution is cheap enough that priority bidding is not the constraint. Ethereum.org's 1 to 6 October 2026 snapshot put Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005 per transaction. Ordering, not inclusion, is the scarce good, and ordering is not sold in the fee market. Q: What would real MEV protection look like? A: A mechanism you can name. Commit-reveal so the operator cannot read the trade before it orders it, encrypted submission with a delay, a published ordering rule applied to every transaction, or several independent parties able to order. A page that says protected without naming one of those is marketing. ## Approval hygiene: the monthly revoke that saves you URL: https://www.caijing188.com/approval-hygiene/ Updated: 2026-10-27T22:09:52+08:00 Verified: 2026-10-27 ### Quick answer An ERC-20 approval outlives the site that requested it, and an unlimited allowance drains without a second signature. Revoking costs what a transaction costs: $2 to $10 on Ethereum L1 in FeeEdge's 17 August 2026 sample, and $0.001 to $0.002 on Base as of October 2026. Review monthly from a spending wallet. ### Key takeaways - An approval stays live after the site closes, changes hands or disappears - Unlimited is an interface default, not a requirement - Permit signatures leave no on-chain record until used, so a revoke sweep does not show them - Revoking twenty approvals costs $2 to $10 each on Ethereum L1 and fractions of a cent on Base, per FeeEdge on 17 August 2026 and Ethereum.org in October 2026 ![A wallet approval list sorted by unlimited allowance](https://img1.ladyww.cn/picture/Picture00454.jpg) An approval is a standing permission written into a token contract, and it outlives the site that asked for it. The site can close, the domain can change hands, the team can disappear, and the allowance stays active until someone sends a transaction to revoke it. Unlimited allowances are the standard drain vector for exactly this reason: one signature, no expiry, no notification. ## What an approval is On EVM chains, an ERC-20 approval writes an allowance into the token contract: a spender address, an amount, and the account that granted it. Once the transaction confirms, the site is no longer part of the arrangement. The contract will honour the spender’s transfer request whenever it arrives, whether that is tomorrow or in two years. That is why “I have not used that site in months” is not a mitigation. Abandoned approvals are the ones that get harvested. ## Unlimited is the default Most interfaces pre-fill the maximum value, because a finite allowance means the user has to approve again later. The convenience is real and so is the exposure: an unlimited allowance to a contract that later turns hostile drains the token without a single further signature from you. The drain usually happens in two steps. A fake mint or claim page gets you to connect and approve, and the attacker waits. Weeks later a transaction sweeps the balance. Nothing on your side changes in between, and nothing warns you. Permit-style signatures deserve their own line. They are signed off chain, cost no gas, and leave no on chain record until they are used. A revoke sweep will not show them, because there is nothing to show yet. ## A monthly review that takes ten minutes Once a month, open a revoke tool, connect the wallet, and sort by allowance size. Then apply three rules. Revoke anything unlimited that you do not actively use. Revoke anything whose spender you cannot identify. Revoke everything attached to a site you have not visited this quarter. Cost is no longer an excuse. A revoke is an ordinary transaction, so it costs what a transaction costs. FeeEdge’s 17 August 2026 sample put an ERC-20 transfer on Ethereum L1 at $2 to $10, which is a real bill for twenty revokes. Ethereum.org’s 1 to 6 October 2026 snapshot put Base at $0.001 to $0.002 and Arbitrum One at $0.004 to $0.005, which makes the same sweep effectively free. If you granted approvals on a cheap chain, clean them there. ## Separate the wallet that signs from the wallet that saves The most effective control is not a tool. It is a second address. Keep savings behind a key that never connects to a site. Fund a spending wallet with roughly a month of activity, and grant approvals only from that wallet. When a new contract appears, use a fresh address rather than the one that already holds permissions, and let the old one run down. The friction is small. Exchange withdrawal fees in FeeEdge’s 17 August 2026 sample were about 0.5 to 1 USDT for Solana SPL, about 0.3 USDT for BEP-20 and highest for ERC-20, so topping up a spending wallet is cheap on most routes. The [hot wallet sizing question](/how-much-in-a-hot-wallet/) covers how much should sit there. | Permission type | How it ends | What you see before signing | What to check | |—|—|—|—| | Finite ERC-20 allowance | Spent down, or revoked | Amount and spender | Whether the spender is still needed | | Unlimited allowance | Only by revoking | Spender, amount shown as unlimited | Anything you cannot name | | Permit signature | Used once, or expires | Rarely rendered clearly | Why a signature needs no gas | | NFT operator approval | Revoked, or the collection is gone | Operator address | Whether it covers the whole collection | | Native transfer | Immediate | Amount and address | Full address, middle characters | ## What it does not cover Revoking does not undo damage. A permit already used, or a balance already swept, stays gone. The review is prevention, and it only works on the wallet you actually audit. Nor does it help with a seed you have exposed. Approvals and key compromise are separate failures: the first is a permission you granted, the second is a secret someone else now has. Only moving the assets addresses the second. ## What to do next Put the review in the calendar for the first weekend of the month, and run it on the wallet that holds the most rather than the one you use most. Ten minutes against an unlimited allowance is the cheapest trade in this column. Then split the roles. One address for sites, one for savings, and a rule that the savings key never sees a connect dialog. If a contract needs funding, it gets a fresh address and a small balance. Finally, treat unlimited as a decision rather than a default. If the interface offers a finite amount, use it. Scoped permissions are also where smart accounts are heading, though [their limits](/account-abstraction-limits/) are real: a policy is only as good as the code enforcing it. ### FAQ Q: Does revoking an approval cost gas? A: Yes, a revoke is an ordinary transaction. In FeeEdge's 17 August 2026 sample an ERC-20 transaction on Ethereum L1 cost $2 to $10, while Ethereum.org's October 2026 snapshot put Base at $0.001 to $0.002. Cleaning up approvals you granted on a cheap chain is nearly free. Q: Can I revoke a permit signature? A: Not directly. A permit is an off-chain signature that leaves no record until a spender uses it, so a revoke tool has nothing to show. The defence is to avoid signing permits you did not initiate, and to keep the wallet that signs them separate from savings. Q: How often should approvals be reviewed? A: Monthly is a workable rhythm, plus after any new site, mint or airdrop claim. Sort by allowance size, revoke anything unlimited you do not use, anything whose spender you cannot name, and anything attached to a site you have not visited this quarter. ## Oracle latency is a liquidation risk, not a data problem URL: https://www.caijing188.com/oracle-latency-liquidations/ Updated: 2026-10-27T22:09:53+08:00 Verified: 2026-10-27 ### Quick answer A lending market liquidates against the last oracle write, not the market price, so a stale feed can make a safe position callable. The deviation threshold matters more than the update interval, and the buffer to liquidation is thin by design. Read the feed configuration, not the market page, as of October 2026. ### Key takeaways - A feed updating every minute can still sit 1.9% from the market on a 2% deviation threshold - The liquidation threshold sits only a few points above max LTV, so the buffer is thin by design - Arbitrum One's 45,818-block dispute window, about 6.4 days, is a settlement clock, not a price clock - Thin liquidator depth turns a stale price into bad debt that suppliers absorb through utilisation ![An oracle feed updating while a lending position sits near its limit](https://img1.ladyww.cn/picture/Picture00435.jpg) A liquidation engine does not know the market price. It knows the last price an oracle wrote to the chain, and it liquidates against that. When the two diverge the difference is not a data quality complaint. It is a transfer of value from a borrower to whoever watches the feed more closely. ## A stale price is a different number, not a wrong one Oracle designs usually combine two triggers. A heartbeat publishes on a schedule, and a deviation threshold publishes early when the price has moved far enough from the last update. Both parameters describe the same thing from different angles, and neither is a promise about the current price. A feed with a twenty-minute heartbeat and a 0.5% deviation threshold will publish steadily through a quiet morning and then hold a price for up to twenty minutes across the one move that matters, because the move happened between heartbeats. The written price is not wrong. It was correct when it was signed. It is a price from the past used as if it were a price from now. ## The threshold matters more than the interval Most checklists ask for the update interval, because the interval is easy to find. The deviation threshold is the parameter that decides how much price movement can accumulate before the contract hears anything. A feed that updates every minute on a 2% deviation can sit 1.9% away from the market indefinitely. A feed that updates every fifteen minutes on a 0.25% deviation tracks far more closely, and it costs more to run because every update is a transaction somebody pays for. That trade-off is why oracle choice is a design decision rather than a procurement one. Cheap feeds on low-volume assets update rarely, and a lending market listing a long tail inherits that thinness. ## How a safe position becomes liquidatable Take a borrower near the maximum loan-to-value, which is where leveraged users sit by construction. The liquidation threshold is a few points above the maximum, so the buffer between opening and being liquidated is deliberately thin. Now let collateral fall 6% in five minutes and let the feed publish it ten minutes later. Two things fill the gap. The borrower sees a health factor that says they are fine, because the dashboard reads the same stale feed. A liquidator running a faster price source sees a position already underwater and can repay the debt and seize collateral at the stale price, which is higher than the market. The liquidator’s profit is the bonus plus the staleness. The borrower’s loss is the same amount, and it is larger than the move that triggered it. The asymmetry runs both ways. A feed lagging a rally overstates collateral, lets borrowers draw more debt than the position supports, and sets up a sharper liquidation when the price snaps back. Latency is biased against the borrower closest to the limit. ## Settlement time is a different clock On a rollup, price freshness and settlement are separate questions. Arbitrum One’s dispute window runs 45,818 blocks, about 6.4 days, per the Ethereum.org L2 directory readings of 1 to 6 October 2026. That is how long a withdrawal takes to become final on the main chain. It says nothing about how fresh the price is on the rollup, and conflating the two is a common error in risk notes. What matters is where the oracle writes and where the liquidation executes. A feed posting on the rollup is fast, and the residual risk is its own threshold. A feed that depends on a main-chain update which must be relayed gives the rollup’s liquidation engine a delay no fee table mentions ([what a sequencer does](/what-a-sequencer-does/)). ## What to check before borrowing near the limit | What to check | Why it matters | Where to look | |—|—|—| | Deviation threshold | Caps how far price can drift before an update | Oracle configuration, not the market page | | Heartbeat | Longest silence in a quiet market | The same configuration | | Liquidation threshold vs max LTV | The real buffer before a call | Market parameters | | Feed count and fallback | One feed is one failure point | Market documentation | | Liquidator depth | Thin liquidation markets mean bigger discounts | Recent liquidation history | The first two rows decide whether the price you see is current. The third decides how much room you have. A market with a stale feed and deep liquidators hurts you precisely. A market with a fresh feed and no liquidators leaves bad debt, and bad debt lands on suppliers through utilisation and rate changes rather than through a single liquidation ([how utilisation sets lending rates](/lending-rates-utilization/)). ## Verdict Oracle latency is a liquidation parameter. It sets the price at which your collateral can be taken, and it is chosen by people optimising cost per update. Borrow well inside the limit, and compute the buffer in units of the feed’s threshold rather than your own reaction time. Your reaction time does not enter the contract. The feed’s settings do. ### FAQ Q: Why does the deviation threshold matter more than the update interval? A: The interval caps how long a feed can stay silent, but the threshold caps how far price can move before an update is triggered at all. A feed with a one-minute interval and a 2% threshold can hold a price 1.9% away from the market indefinitely, which is the gap that matters near a liquidation boundary. Q: Can a stale oracle hurt a borrower when the price moves in their favour? A: Yes. A feed lagging a rally overstates collateral, which lets borrowers draw more debt than the position supports. When the price snaps back and the feed catches up, the liquidation is sharper than it would have been with a fresh price. Q: What should be checked before borrowing near the maximum LTV? A: The deviation threshold and heartbeat of the feed, the gap between liquidation threshold and max LTV, how many feeds back the market and whether there is a fallback, and how deep the liquidator market has been in recent stress. The first two say whether the price is current and how much room exists. ## Stablecoin depeg: what actually happens in the first hour URL: https://www.caijing188.com/stablecoin-depeg-first-hour/ Updated: 2026-10-27T22:09:53+08:00 Verified: 2026-10-27 ### Quick answer In the first hour of a depeg, redemption queues and thin books matter more than the last print. Stablecoin supply was $292.40B as of October 2026 per RWA.xyz, and Base or OP Mainnet exits take seven days per Ethereum.org's 1 to 6 October 2026 snapshot. Check chain and venue before stress. ### Key takeaways - The first hour produces two prices: a thin on-chain print and a redemption value that clears later - Ethereum.org's 1 to 6 October 2026 snapshot lists seven-day exits for Base and OP Mainnet and 45,818 blocks, about 6.4 days, for Arbitrum - Stablecoin supply of $292.40B in October 2026 per RWA.xyz is a large float to reallocate under stress - Most retail holders cannot redeem with the issuer, so the venue's book is their only exit ![A stablecoin price chart gapping below one dollar during a stress event](https://img1.ladyww.cn/picture/Picture00405.jpg) A stablecoin depeg is usually described as a price event. In the first hour it is closer to a queueing event. The price on the screen is the last trade that cleared, and the books behind it are thin, wide and repricing faster than any redemption can settle. Stablecoin supply reached $292.40B in October 2026, per RWA.xyz, which is a large float to reallocate in a hurry. ## The first hour is a queue, not a price In the first minutes of a stress event, sellers arrive together and makers widen or pull. The quoted price becomes a poor estimate of what a large order would realise, because the depth that produced the quote is gone. Redemption is a separate queue with a separate clock. Issuers process redemptions in business hours, through banking partners, with their own cutoffs and minimums. A holder who submits at 09:00 on a stress day is behind everyone who submitted the night before, and the settlement is in fiat, which takes its own time. So the market carries two prices in the first hour: a thin on-chain price that moves immediately, and a redemption value that will not clear until later. The gap between them is not a forecast. It is the cost of waiting, and anyone pricing the event off the last trade is pricing the queue’s impatience rather than the issuer’s balance sheet. ## On-chain pools reprice before fiat clears An automated market maker pool reprices on the next trade. There is no queue and no cutoff, only the curve and the inventory in it. That is why the on-chain price breaks first and overshoots first. The overshoot is structural. A pool with limited inventory cannot absorb a large redemption without moving far along its curve, and the movement is largest when several sellers arrive together. The price then mean-reverts as arbitrageurs bring inventory in, if they can. Fiat redemption is the opposite. It is slow, documented, and usually reliable for holders who qualify. It clears at par minus fees, days later. That is fine for a treasury and useless for a trader watching a mark. ## Which chain holds your float The question to answer before stress is where the tokens sit. Not which issuer, which chain. Exit speed varies by orders of magnitude. The Ethereum.org L2 directory, in its 1 to 6 October 2026 snapshot, lists seven-day mainnet withdrawals for Base and OP Mainnet, a dispute window of 45,818 blocks, roughly 6.4 days, for Arbitrum, and a minimum of three hours for ZKsync Era. A stablecoin on a seven-day exit is not a stablecoin you can redeploy inside a first hour. Bridge cost is a second tax. FeeEdge’s survey of 17 August 2026 put Solana transfers below $0.01, Polygon PoS at $0.002 to $0.01, BNB Chain at $0.01 to $0.20, Base below $0.01 to $0.10, Arbitrum at $0.01 to $0.50, and Tron at $1 to $3. ## Which venue can redeem, and for whom Issuers redeem for qualifying holders, typically institutions and registered accounts, with minimums that exclude most retail balances. A retail holder’s practical redemption is a sale on a venue, at whatever the book pays. That makes venue choice part of the risk. A venue with a direct issuer relationship and a fiat rail is a different asset from a venue with neither, even when both show the same ticker. | Where the float sits | First-hour behaviour | What to check in advance | |—|—|—| | On a venue with a fiat rail | Sale at book price, then a queue for fiat | Whether your account tier qualifies | | On a seven-day L2 | Cannot move during the first hour | Which chain the balance sits on | | In an on-chain pool | Reprices at once, can overshoot | Pool inventory and depth | | On a fast chain | Movable within minutes | Bridge and withdrawal fees | | In an issuer account | Redeems at par, clears later | Cutoff times and minimums | ## What to check before stress The first hour is the wrong time to learn any of this. The checks are cheap in calm markets and impossible in a queue. Confirm which chain each balance sits on, and what the exit from that chain costs in time and fees. Confirm which venue can turn the token into fiat for an account like yours, and what the withdrawal limit is. Confirm whether the issuer would redeem directly, which for most retail holders it will not. Write the numbers down while the market is calm. The [stablecoin supply breakdown](/stablecoin-supply-who-holds-it/) covers how much of the float sits where, and the [lending market data](/lending-rates-utilization/) shows how fast utilisation moves when holders borrow against the same collateral. ## Verdict In the first hour, the price is the least informative number on the screen. The queue is the real market: who can redeem, on which chain, at what cutoff. Check the chain, the venue and the redemption terms before stress, not during it. A stablecoin is only as stable as the fastest path from the token to the currency you actually need. ### FAQ Q: Why does the on-chain price fall before redemption terms change? A: Automated market maker pools reprice on the next trade, with no cutoff and no queue. A pool with limited inventory must move far along its curve to absorb a large sale, so the print overshoots before arbitrage inventory arrives. Fiat redemption still clears at par for qualifying holders, days later. Q: Does the chain a stablecoin sits on matter during a depeg? A: It sets the exit clock. Ethereum.org's 1 to 6 October 2026 snapshot lists seven-day mainnet withdrawals for Base and OP Mainnet, a dispute window of 45,818 blocks, roughly 6.4 days, for Arbitrum, and a minimum of three hours for ZKsync Era. A seven-day exit cannot be used inside a first hour. Q: What should be checked before any stress event? A: Which chain each balance sits on, what the exit costs in time and fees, which venue can convert the token to fiat for an account like yours, and what the withdrawal limit is. All four are cheap to confirm in calm markets and impossible to confirm in a queue. ## The travel rule: what actually travels with your transfer URL: https://www.caijing188.com/travel-rule-in-practice/ Updated: 2026-10-27T22:09:54+08:00 Verified: 2026-10-27 ### Quick answer Travel rule messages carry originator and beneficiary data between regulated venues, which is why a self-custody withdrawal can trigger extra questions, as of October 2026. Our board ranks 24 venues, and an ERC-20 withdrawal costs $2 to $10 (FeeEdge, 17 August 2026). A not supported message usually describes the sending venue's own list. ### Key takeaways - Originator and beneficiary data travel with the payment between regulated venues - A self-custody withdrawal leaves the beneficiary fields thin, which is why it triggers questions - Not supported describes the sending venue's list today, not a legal status - Keep the venue record of each transfer: amount, asset, network, fee and address, dated ![A transfer confirmation showing sender and beneficiary fields](https://img1.ladyww.cn/picture/Picture00425.jpg) When two regulated venues move funds between each other, a data message usually travels with the payment. It names the sender, the recipient and the account on each side. That is the travel rule in practice, and it is the reason a withdrawal can be delayed by a question that has nothing to do with the blockchain. ## What the rule asks for Frameworks that implement the travel rule require the sending institution to pass originator and beneficiary information to the receiving institution, and to keep a copy. The thresholds, the exact fields and the treatment of self-hosted wallets differ by jurisdiction, so treat any specific number you read as local rather than universal. The principle is constant. The payment and the identity travel together, so the receiving venue can decide whether to accept the funds before they are credited. That is a screening decision made at the door, not an audit after the fact. ## What travels, field by field ### Originator Name, account number or address, and often the customer’s address. Some regimes add a national identity number or date and place of birth. The sending venue already holds this from onboarding, which is why supplying it is usually instant. The weak field is the address. An account number at a venue is verifiable. A self-custody address is a string, and its ownership is asserted rather than proven. ### Beneficiary Name and account or address. Where the beneficiary is another venue’s customer, the receiving venue can match the name to its own file. Where the beneficiary is a self-hosted wallet, the fields are thinner and there is less to match against. That gap is not a loophole. It is the place where manual review happens. ## Why a self-custody withdrawal gets questions A transfer to your own wallet is a payment to an unnamed counterparty from the venue’s point of view. The venue cannot verify who controls the address, so it may ask for proof of ownership, a screenshot of the wallet, a signed message, or a short statement about the purpose. This is usually a file that needs an answer before release, not a suspicion. The practical cost is time, and the practical fix is consistency. A labelled withdrawal address, used repeatedly, gives the venue a history to look at instead of a one-off question. Address hygiene matters here too. Our note on [address poisoning](/address-poisoning-habit/) covers the copy-paste failure that sends a correct payment to a lookalike destination, which then needs a very different conversation. ## What “not supported” means A message saying a jurisdiction or a destination is not supported usually describes the sending venue’s own list. Venues maintain country lists, counterparty lists and product lists, and they revise them without a public changelog. Read the message literally. It says this venue will not send there, today, on this rail. It does not say the destination is unlawful, and it does not say another venue would refuse. The useful follow-up is to ask which list the decision came from and whether a different rail changes it. Sometimes the answer is a network restriction rather than a country one. ## The trail you keep Save the withdrawal record at the time of transfer: amount, asset, network, fee, address and date. An explorer link is useful, but it does not name you. A venue record does. That record is also what a receiving venue will ask for if it holds the deposit. Producing it in ten minutes is the difference between a delay and a week. | Field | Usually collected by | What happens if it is missing | |—|—|—| | Originator name | Sending venue, at onboarding | Transfer held until the field is supplied | | Originator account or address | Sending venue, per transaction | Receiving venue cannot match the payment | | Beneficiary name | You, at the withdrawal screen | Held for clarification | | Beneficiary address | You, per transaction | Extra ownership evidence requested | | Purpose or relationship | Sometimes, for self-hosted wallets | Manual review, with no fixed timeline | ## What to do next Keep one labelled withdrawal address per purpose, and use it consistently. Keep the venue receipt for each transfer, not only the explorer hash. If a transfer is held, answer with the record rather than with an explanation. Cost is the secondary issue. An ERC-20 withdrawal ran $2 to $10 in FeeEdge’s 17 August 2026 survey, above $50 in congestion, while a Solana SPL withdrawal was about 0.5 to 1 USDT. A held transfer costs more than either, in time if not in fees. The rule does not change what a chain records. It changes what a venue must know before it credits you, and that is the part you can prepare for. The [venue board](/exchange/) lists 24 venues as of October 2026, and the [methodology note](/methodology/) explains how the columns are built. ### FAQ Q: What information travels with an international crypto transfer? A: Typically the originator's name and account or address, the beneficiary's name and account or address, and enough context for the receiving venue to identify the payment. Some jurisdictions add identity numbers or dates of birth. Thresholds and required fields differ, so treat any specific figure as local rather than universal. Q: Why is a withdrawal to my own wallet questioned? A: Because from the venue's side the beneficiary is an unnamed party controlling an address it cannot verify. The fields that would normally carry a name are empty or self-declared. The venue may ask for proof of ownership, a signature or a statement of purpose before releasing the transfer. It is usually a file completion, not an accusation. Q: What does a not supported jurisdiction message mean? A: It means the sending venue will not send to that destination today, on that rail, under its own policy. Venues maintain country and counterparty lists and update them without a public changelog. The message does not establish that the destination is unlawful, and it does not predict what another venue would do. ## What a node actually verifies for you URL: https://www.caijing188.com/what-a-node-verifies/ Updated: 2026-10-27T22:09:54+08:00 Verified: 2026-10-27 ### Quick answer A light client, a full node and a hosted RPC endpoint answer different questions, and only the first two check the chain. Ethereum.org listed, as of October 2026, Base fees at $0.001 to $0.002 and Linea near $0.024, low enough that most users never question the endpoint. Balances and nonces can be verified; old history usually cannot. ### Key takeaways - A full node recomputes state, a light client verifies proofs, and an RPC endpoint answers questions - Ethereum.org's 1 to 6 October 2026 snapshot put Base at $0.001 to $0.002 per transaction - Balances and nonces are checkable by proof, while historical queries usually are not - FeeEdge's 17 August 2026 survey put L1 ERC-20 withdrawals at $2 to $10, above $50 under congestion ![A light client, a full node and a hosted endpoint side by side](https://img1.ladyww.cn/picture/Picture00415.jpg) Verified on-chain is the most overused phrase in wallet interfaces. What is verified depends on which of three very different things your software is talking to: a light client, a full node, or a remote endpoint run by someone else. Two of those check the chain. The third asks a server. ## Three machines, three questions A full node downloads and executes blocks itself. It checks that each transaction follows the rules, that balances add up, and that the block’s state root matches the state it computed. When it reports a balance, that balance is a fact the node established rather than one it was told. A light client does not execute everything. It follows block headers and requests proofs about specific pieces of state, checking each proof against a root it trusts. It stores far less than a full node. An RPC endpoint answers queries. Behind it may sit a full node, an archive node, an indexer or a cache, and the operator may be entirely honest. The structural point is that the answer arrives as a response, not as a proof your software checked. ## What a hosted endpoint really gives you A hosted RPC is an indexer with an API. It stores decoded state, answers balance and nonce questions from its own tables, and serves historical queries from a database it built by replaying the chain. That design is why it is fast and why it is useful. It is also why the answers are not self-certifying. When your wallet asks for a balance it receives a number with no accompanying proof, and the wallet cannot tell a fresh reading from a stale one. Most of the time this is invisible: the endpoint is accurate and the stakes are small. Ethereum.org’s L2 directory, in its 1 to 6 October 2026 snapshot, put fees as low as $0.001 to $0.002 per transaction on Base, with Arbitrum One at $0.004 to $0.005 and Linea near $0.024. At those prices nobody audits the source of a balance. ## How to tell which one you are on The interface rarely says. Three habits settle it. Look for a sync indicator. A full node reports block height and peers, and its height advances with the chain. A hosted endpoint may show a height too, but the vendor supplied it. Look for a proof. Light clients verify state proofs, and some wallets say they did. If nothing in the flow mentions a proof or a header chain, you are reading an answer. Look at the URL. A localhost or LAN address is your machine. A vendor domain with an API key is someone else’s, and that is custody of information rather than of funds. | Question | Who can answer it verifiably | What fails quietly on a hosted endpoint | |—|—|—| | What is my balance | A full node, or a light client with a state proof | A cache that has not refreshed | | What is my nonce | A full node | An indexer lagging the tip | | Did my transaction land | A full node following the canonical chain | An endpoint on a different fork tip | | What did this address do years ago | An archive node you run | Pruned history replaced by a vendor database | | What is my token balance | A node plus the token contract’s own state | The indexer’s own accounting of transfers | ## When it matters It matters most where an answer decides whether a transaction is valid. A balance and a nonce are inputs to signing. If either is wrong, the transaction fails, gets replaced, or gets built on state that does not exist. It matters least for historical queries, where the cost of being wrong is a wrong chart rather than a wrong signature, and where almost nobody runs the archive node to check. In between sits everything that touches money at the margin: whether a withdrawal has settled, whether a contract call reverted, whether an allowance is still live. Those answers are checkable, and they are the ones a hosted endpoint is most likely to serve from a cache. Withdrawal costs make the same point from the other side. FeeEdge’s survey of 17 August 2026 put L1 ERC-20 withdrawals at $2 to $10, and above $50 under congestion, while exchange withdrawal fees ran about 0.5 to 1 USDT for Solana SPL and about 0.3 USDT for BEP-20. The expensive leg is the one people check twice. The cheap leg is the one they trust. ## What to do next Decide which questions you are outsourcing, then stop outsourcing the two that decide validity. If you run nothing, prefer a wallet that verifies proofs for balance and nonce, and treat every other number as a vendor’s opinion. If you move size, run a node for the chain you settle on, or use an endpoint whose operator you could name in a dispute. Read [the limits of account abstraction](/account-abstraction-limits/) for where the signature layer stops helping, and the [wallets section](/category/wallets/) for how different clients handle state. An endpoint is a convenience. A node is a position, and the difference shows up on the day the two disagree. ### FAQ Q: If my wallet shows a balance from an RPC, is it verified? A: No. The endpoint returns a number and your wallet displays it. Unless the wallet verifies a proof against a block header it trusts, the balance is an answer from a server. A full node, or a light client with a proof, turns that answer into something the software checked, as of October 2026. Q: Do I need to run a full node to be safe? A: Not for every action, but you need to know which questions you are outsourcing. Balances and nonces are worth verifying because they decide whether a transaction is valid. Historical and indexed queries, such as a full token history, are the ones almost nobody verifies, hosted or not. Q: Why does a hosted endpoint sometimes disagree with a node? A: Because it may be answering from a cache, a lagging indexer, or a different fork tip. None of those are necessarily dishonest. They are different data sources answering a question that sounds identical, and the interface does not tell you which one replied. ## Blind signing: why the screen is the only signature you can read URL: https://www.caijing188.com/blind-signing-screen/ Updated: 2026-10-27T22:09:55+08:00 Verified: 2026-10-27 ### Quick answer When a wallet cannot decode a payload it shows a hash and asks for approval, and a hash cannot be checked against intent. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026. The fix is a screen that renders the amount and the full destination, plus a policy of refusing anything else. ### Key takeaways - A digest cannot be compared with an intention, only with another digest - One blind-signed payload can carry an unlimited approval, an operator approval or a proxy call - A usable screen answers three questions: how much, to whom, which function - Strict decoding turns an unreadable request into a stop, at the cost of failed edge cases ![Device screen showing a raw hash instead of a decoded amount](https://img1.ladyww.cn/picture/Picture00455.jpg) A signature is a statement, but only if you can read it. When a wallet cannot decode a payload it falls back to showing a hash and asks for approval, and a hash cannot be checked against intent. That fallback is the quiet path around every other control in self custody: the key stays in the secure element, the device is genuine, the firmware is signed, and the transaction still sends everything to an address you have never seen. ## What the device is actually asked to do Signing produces a signature over a digest. The digest is a fixed length value derived from the message, and the device’s real work is deciding what the message says before it signs. Firmware carries decoders for common payloads. A plain transfer, a known token contract, a familiar staking call: each has a template, so the screen can show an amount and an address. When the payload matches no template, the device has two options. It can refuse, or it can show the digest and let you decide. ## A hash is not consent A hash is a fingerprint, not a sentence. You can compare it with another hash, which is useful when you built the transaction yourself on an offline machine and can check both ends. You cannot compare it with an intention. Consider what a blind signed digest can carry. An unlimited token approval. An operator approval over an entire NFT collection. A permit that lets a third party move a balance at any time. A call to a proxy that forwards to something else. None of that is visible in a digest, and each one is irreversible once the transaction confirms. The mental trap is the word approve. Pressing a button labelled approve on a device feels like approving a specific action. It is approving a signature over bytes you have not read. ## Why the wallet cannot always decode Decoding is hard because calldata is open ended. Any contract can define any function, and new contracts appear faster than firmware updates. Proxies and delegate calls add a layer: the call names one contract and executes another. There is a legitimate use for the fallback. Recovery flows, unusual script paths and some multisig coordination produce payloads a consumer device will never have a template for. Those are exactly the cases where the user built the transaction deliberately and can verify the digest against a second source. The illegitimate use is everything else, and it is the common case. A hostile site does not need a novel exploit. It needs a payload the device cannot render. ## The screen test Three questions decide whether a wallet is usable for signing. Does it show the amount in the asset, with the right decimals? Does it show the full destination, including the middle characters? Does it name the function when the payload is not a plain transfer? If the answer to all three is yes, the device is doing the only job that matters at the moment of signing. If any answer is no, the hardware is protecting a key you are about to spend without reading the instruction. Truncated addresses are the place this bites most often, and the habit that catches them is in [the address poisoning note](/address-poisoning-habit/). Chain choice changes the stakes rather than the principle. Ethereum.org’s L2 directory put Base at $14.6B of value in October 2026, and every position in that figure moves by a signature someone had to read. | Payload | Can a screen render it | What a hash-only fallback hides | |—|—|—| | Native transfer | Yes | Amount and destination | | Token transfer | Yes, with decimals | Which token contract | | Unlimited approval | Yes, if the wallet decodes approve | The spender, and the word unlimited | | Permit signature | Often not | That no gas is being spent | | Proxy call | Sometimes | The contract that actually executes | | Batch or multicall | Rarely | Every action inside the batch | ## Refusing is a control The strongest setting is a device that will not sign an undecoded payload at all. The setting exists on some devices and is off by default on many, because it breaks edge cases. Turning it on converts an unreadable request into a stop. The transaction fails, you go and find out what it was, and the default answer becomes no. For a savings device that is the right default. For a device used daily the friction may be too high, which is an argument for separating the two. ## What to do next Enable strict decoding on the device that holds savings, and accept that some transactions will fail. Keep a second, smaller setup for the experiments that need the fallback. Read the screen, not the companion app. The app is software on a general purpose computer, and the device exists because that computer is not trustworthy. If the two disagree, the device is right and the transaction is wrong. Then re-test the habit after every firmware update. Decoders change, defaults change, and a setting enabled two years ago may not have survived. The [limits of account abstraction](/account-abstraction-limits/) explain why a smarter account does not remove this step. ### FAQ Q: Is blind signing ever legitimate? A: Yes, for payloads you built yourself and can verify against a second source, such as recovery flows, some script paths and some multisig coordination. The problem is the default, where an undecodable request from a website is treated as routine. On a savings device the safe setting is refusal. Q: Why can a hardware wallet not decode every transaction? A: Calldata is open ended: any contract can define any function, and new contracts appear faster than firmware updates. Proxies and delegate calls add a layer where the named contract is not the one that executes. Decoders cover common patterns, and the rest falls back to a digest. Q: Does a bigger or newer device fix this? A: Not by itself. What matters is whether the firmware carries decoders for the payloads you actually sign and whether the screen has room to show a full address and an amount with the right decimals. A device that renders raw units has moved the arithmetic back to you. ## Looping: where the leverage hides inside a four percent spread URL: https://www.caijing188.com/looping-hidden-leverage/ Updated: 2026-10-27T22:09:55+08:00 Verified: 2026-10-27 ### Quick answer Looping turns a four point lending spread into a leveraged position: supply, borrow, supply again. Three steps on an 80% loan-to-value gives 2.44 times leverage and a health factor of 1.40, leaving about 28.5% of collateral fall before liquidation, as of October 2026. The spread sizes the trade; the health factor decides whether you keep it. ### Key takeaways - Three loop steps at an 80% LTV give 2.44 times leverage and a 59% blended LTV - At 2,440 supplied and 1,440 borrowed, an 82.5% liquidation threshold allows about a 28.5% fall - Carry at an assumed 8% supply and 4% borrow is 137.6 on 1,000 of equity, or 13.8% - Looping adds no capital, so every extra point of carry is bought with borrowed money ![A looped supply and borrow position drawn as stacked layers](https://img1.ladyww.cn/picture/Picture00436.jpg) A four point spread between what a market pays suppliers and what it charges borrowers looks like a modest carry trade. Looped through the same market it becomes a leveraged position whose outcome depends on a health factor, and the leverage is invisible on any page showing the spread instead of the balance sheet. ## The loop, step by step Start with 1,000 USDC in a market with an 80% maximum loan-to-value. Supply the 1,000. Borrow 800 against it. Supply the 800. Borrow 640. Supply the 640. After three steps the position supplies 2,440 and owes 1,440. Equity is still 1,000, because every borrowed unit went straight back into collateral. Leverage is 2.44 times. The loan-to-value on the whole position is 59%, not 80%, because only the final tranche sits at the maximum. Keep looping and the average climbs toward 80%: the series converges, so each step is smaller than the last and the position approaches the limit without reaching it. That convergence is why “loop five times” and “loop to the maximum” are nearly the same trade. Risk per step shrinks and the risk already on the book does not. ## The spread is not the yield Carry is supply multiplied by the supply rate, minus debt multiplied by the borrow rate. Two things drive it: the spread on borrowed money, and the base rate on your own. Run it with a stated assumption, not a market quote: supply at 8%, borrow at 4%, 2,440 supplied, 1,440 borrowed. Gross carry is 195.2 minus 57.6, or 137.6 on 1,000 of equity, about 13.8%. Change the assumption and the intuition breaks. With supply at 5% and borrow at 1% the same loop earns 122 minus 14.4, or 10.8% on equity, for the same four point spread. With supply at 8% and borrow at 7% the carry falls to 9.4%. | Item | Amount | Note | |—|—|—| | Equity | 1,000 | Unchanged by looping | | Supplied | 2,440 | Three supply steps | | Borrowed | 1,440 | Three borrow steps | | Leverage | 2.44x | Supplied divided by equity | | Carry at 8% and 4% | 137.6 | Assumption, not a market quote | Looping adds no capital, so every extra point of carry is bought with borrowed money and paid for in liquidation risk. ## The health factor is the whole trade The health factor is collateral value times the liquidation threshold, divided by debt. Above one, nothing happens. Below one, anyone may repay part of the debt and take collateral at a discount. At 2,440 supplied, 1,440 borrowed and an 82.5% liquidation threshold, the health factor is 1.40. Collateral can fall about 28.5% before the position is callable. That reads as comfortable, and it is the number a dashboard shows you. Two things make it the wrong number to lean on. The buffer narrows as steps are added, because the average LTV converges upward toward the maximum. And it is measured against a price, so a fall arriving in one candle does not come with 28.5% of time to react. It produces a liquidation and a bonus paid out of your collateral. ## The exit you cannot take Unwinding means repaying debt and withdrawing collateral, both in the same market, and in a crunch the market is the problem. As utilisation approaches its ceiling the withdrawable balance shrinks, because most of the supplied asset is out on loan. A looped position that wants to deleverage needs exactly that balance, at the same moment as every other looped position. This is where the comparison with a futures market breaks. A futures position can be closed against a book that clears at a price. A lending position can only be closed if the pool holds unused liquidity, and the pool is least liquid when the trade is most crowded ([what utilisation does to lending rates](/lending-rates-utilization/)). There is a worse exit. Selling collateral to repay debt works only if the collateral is the borrowed asset. A position looped with a volatile collateral against a stablecoin borrow cannot be deleveraged in slices, because the sale moves the price the health factor depends on. Part of the exposure is operational too: a loop run from a hot wallet is one key from a total loss ([how much sits in a hot wallet](/how-much-in-a-hot-wallet/)). ## What to do next Before the first loop, write down three numbers: the health factor at entry, the collateral fall that takes it to one, and the pool’s withdrawable liquidity at that instant. Then re-run the third assuming half the other looped positions want out at the same time. If the position only works at full leverage, it does not work. It is a bet on the spread holding, the price holding, and the pool staying liquid. ## Verdict Looping is not a yield strategy. It is a leveraged long position in the spread between two rates, collateralised by an asset whose price you do not control, financed by a pool whose liquidity you share with everyone running the same trade. The four point spread is the reason to look. The health factor is the reason to size. The withdrawable liquidity is the reason the exit is priced by somebody else. ### FAQ Q: Is looping the same as leverage? A: Yes, with different collateral and a different liquidation mechanism. The borrowed asset is re-supplied, so position size grows while equity stays fixed. Leverage equals total supplied divided by equity, which is 2.44 times after three steps in the example here. Q: Why is the blended loan-to-value lower than the maximum? A: Because only the last tranche is borrowed at the maximum. Earlier tranches were borrowed against collateral that has since been counted again, so total debt over total supply sits below the cap. Adding steps pushes the blended figure toward the maximum, which is why loop counts converge. Q: What breaks a looped position first, price or liquidity? A: Price sets the health factor, so a sharp collateral fall is what triggers a liquidation. Liquidity determines whether you can deleverage before that happens. In practice the two arrive together, because a collateral fall usually accompanies the utilisation spike that removes the withdrawable balance. ## Tokenised funds: the licensing question product pages skip URL: https://www.caijing188.com/tokenised-fund-licensing/ Updated: 2026-10-27T22:09:56+08:00 Verified: 2026-10-27 ### Quick answer A token can be a fund unit, a debt claim or a receipt, and that label decides who may sell it and to whom, as of October 2026. Tokenised Treasury products held $14.64B to $14.90B across 117 funds (RWA.xyz), and most require KYC to mint or redeem. Check the label before the yield. ### Key takeaways - Tokenised Treasury products held $14.64B to $14.90B across 117 funds in October 2026 (RWA.xyz) - Distributed RWA value was $39.01B, up from about $21.5B at the start of 2026, a rise of 81% - Of $34.865B in on-chain AUM, only $4.182B was deployed into DeFi (DefiLlama) - Most tokenised Treasury products require identity verification to mint or redeem ![A fund factsheet beside a token contract address on a screen](https://img1.ladyww.cn/picture/Picture00426.jpg) Every tokenised fund page shows a yield and a chain. Few show the line that decides whether you can hold it: what the token legally is. A fund unit, a debt claim and a receipt are three different instruments, and the label determines who may sell the product, to whom, and under what disclosure. ## The label is the licence The label is not marketing. A fund unit usually sits inside a collective investment regime, which brings a prospectus, a depositary and a restriction on who may be sold to. A debt claim is a security issued by a vehicle under an offering document. A receipt is a claim on an intermediary, and its treatment follows the intermediary. The same yield can be packaged three ways. The yields converge. The rules do not. That is why two products quoting the same Treasury rate can have completely different investor bases, and why one can be sold on a public venue while the other cannot. ## Three labels, three rulebooks ### Fund unit The holder owns a share of a pooled portfolio. The fund has a manager, a custodian and a document that states the investment policy. Sales are restricted by the regime the fund sits in, which usually means a distributor authorised for that regime and a buyer who meets the criteria. ### Debt claim The holder owns a contractual claim on an issuer or a special purpose vehicle. There is an offering document rather than a prospectus, and the terms set out the maturity, the coupon and the ranking. Transfer restrictions are contractual, and they bind whoever holds the token. ### Receipt The holder owns a claim on an intermediary that holds the asset. The token is a record of that claim, not the asset itself. The risk is the intermediary, and the rules that apply are the ones governing the intermediary’s custody arrangement. ## Why KYC sits at the mint Most tokenised Treasury products require identity verification to mint or redeem, as of October 2026. That is the visible edge of the label. The primary market is gated because the wrapper is a regulated instrument and the issuer has obligations to the buyer. The secondary market is where the label gets tested. A token that trades freely on a venue may be trading outside the terms of the wrapper, and the issuer’s transfer restrictions tell you which reading applies. If the document says transfers require an eligible holder, a venue listing does not override it. Read the transfer section before the yield section. It is usually shorter and it decides more. ## What 117 funds tell you Tokenised Treasury products held $14.64B to $14.90B across 117 funds in October 2026 (RWA.xyz). The whole distributed RWA market was $39.01B, up from about $21.5B at the start of the year, a rise of 81%. Private credit was about $8B, commodities above $5B, tokenised equities about $3.3B and real estate about $226M, under 1% of the total. Concentration matters as much as size. Ethereum carried $16.05B of distributed value, Securitize $4.38B and Ondo $3.96B. Of $34.865B in on-chain AUM counted by DefiLlama, only $4.182B was deployed into DeFi, with Morpho Blue at $735.9M and Aave V3 at $437.6M. The gap between the two figures is the point. Most of this market is held in wallets and permissioned accounts rather than lent out, which is what a wrapper with transfer restrictions looks like from the outside. | Label on the token | What the holder owns | Who may sell it, in general terms | |—|—|—| | Fund unit | A share of a pooled portfolio | A distributor authorised under the fund’s regime, to buyers who meet its criteria | | Debt claim | A contractual claim on an issuer or vehicle | The issuer or an arranger acting under the offering document | | Receipt | A claim on an intermediary holding the asset | The intermediary, under its custody terms | | Feeder or wrapper share | An interest in another vehicle that holds the underlying | Whoever the wrapper’s rules permit, which can be narrower than the underlying fund | ## What to do next Find the noun. In the offering document, the terms or the risk section, look for share, note, receipt or interest. Then check three things against it: who the issuer is, which regime the document references, and whether transfers are restricted. If the page cannot tell you, the yield is not the number to compare. Our breakdown of [the three products inside the $39B](/thirty-nine-billion-three-products/) sorts the market by wrapper, and [lending rates and utilisation](/lending-rates-utilization/) explains what happens when a wrapper is deployed rather than held. A product that will not name its label has not skipped a detail. It has skipped the question that decides whether it can be sold to you at all. ### FAQ Q: Why does the label on a token matter to an investor? A: Because the label determines the rulebook. A fund unit sits inside a collective investment regime with a prospectus and restrictions on who may be sold to. A debt claim is issued under an offering document. A receipt is a claim on an intermediary. The same yield can be wrapped three ways, and only one of them may be sellable to you. Q: Why do most tokenised Treasury products require KYC? A: Because the wrapper is a regulated instrument and the issuer has obligations to the buyer, so the primary market is gated at the mint. A transferable token can still move on a chain, but that does not make the wrapper unrestricted. The issuer's transfer rules tell you which reading applies. Q: Does a large tokenised Treasury market mean the tokens are widely traded? A: Not necessarily. Of $34.865B in on-chain AUM counted by DefiLlama in October 2026, only $4.182B was deployed into DeFi, with Morpho Blue at $735.9M and Aave V3 at $437.6M. The rest sat in wallets and permissioned accounts, which is consistent with wrappers that are held rather than traded. ## Passphrase, second seed, or split: three ways to survive a loss URL: https://www.caijing188.com/second-factor-seeds/ Updated: 2026-10-27T22:09:57+08:00 Verified: 2026-10-27 ### Quick answer A BIP39 passphrase defeats a thief who finds the seed, but a forgotten passphrase is total loss, a second seed doubles the bookkeeping, and splitting trades convenience for recovery risk. The stablecoin float was $292.40B as of October 2026, per this board's data pack. Match the scheme to the event you fear. ### Key takeaways - A passphrase stored beside the seed is one factor with extra steps - A forgotten passphrase produces no error, only a different empty wallet - Shamir 2-of-3 survives one lost location and one stolen share; split halves survive neither - None of the three stop a signature you approved, which is the failure that needs no seed ![Three backup layouts: passphrase card, second seed, split shares](https://img1.ladyww.cn/picture/Picture00456.jpg) A single seed is one object that must never be lost and never be found. Every scheme for surviving a loss is an attempt to break that pairing, and each one adds a new way to fail. A BIP39 passphrase, a second independent seed, and a split backup protect against different events, and the comparison only makes sense once the event is named. ## The problem being solved Two risks pull in opposite directions. Loss risk says make copies; theft risk says every copy is a target. A copy in a second city adds a place a burglar can look. Any scheme answers three questions. What happens if a location burns? What if someone finds one copy? And what new single point of failure did it create? ## Passphrase: a factor, not a copy A BIP39 passphrase is an extra word or phrase mixed into key derivation. The same 24 words with a different passphrase produce a different wallet, with no error message and no way to tell that a wallet exists behind it. That property is the point. A thief who finds the seed and not the passphrase sees an empty wallet. The trap is symmetric: a forgotten passphrase is indistinguishable from a wrong one, and the funds behind it are unrecoverable. The failure to avoid is storing the passphrase the same way as the seed. Written on the same card, saved in the same password manager, known to the same person: each of those collapses two factors into one. The passphrase adds protection only when it lives where the seed does not, and when at least one other person knows it exists. ## A second seed doubles the surface Two independent wallets, each with its own seed and its own backup, is the simplest scheme to operate. Nothing is shared, so no single discovery compromises both, and no derivation trick is required. What doubles is the bookkeeping. Two seeds, two backup procedures, two sets of locations, two things to test. The failure that appears is administrative: one of the two is backed up properly and the other is not, and nobody notices until the day it matters. Used deliberately, the second seed is a separation tool rather than a backup. Savings live behind a seed that never touches a browser; spending lives behind one expected to be exposed. That separation is what earns the complexity, and it is the [hot wallet split](/how-much-in-a-hot-wallet/) at a larger scale. ## Splitting trades convenience for recovery risk Splitting one seed into shares removes the single object. It also creates a reconstruction step, and reconstruction is where schemes fail, because it happens once, under stress, possibly years after setup. Shamir style sharing with a threshold is the robust version. Three shares, any two sufficient, is the usual configuration: no location holds a secret, one location can be lost, and one share can be stolen without consequence. It needs a generator you trust and a written record of the threshold, because a share without its siblings is a random string. Halves of a word list are the crude version. Twelve words here and twelve there cannot be combined without both, and no software is needed. The trade is availability: lose either half and the wallet is gone. ## What none of them cover None of these protect against a signature you approved. A drainer needs no seed, and an unlimited approval empties a wallet that is backed up perfectly. That failure belongs to approval review and to [the address poisoning habit](/address-poisoning-habit/), not to backup design. None protect against coercion. A person who knows you hold assets can ask for the passphrase as easily as the seed. | Method | Stops a thief who has the seed | Survives one lost location | New single point of failure | |—|—|—|—| | Single seed, two copies | No | Yes | One discovery ends both | | Seed and passphrase, stored apart | Yes | Yes, if both survive | A forgotten passphrase | | Seed and passphrase, stored together | No | No | One find ends both | | Two independent seeds | Partly, one wallet only | Yes | One wallet backed up badly | | Shamir 2-of-3 | Yes | Yes | Losing two shares | | Split halves | Yes, from one half | No | Losing either half | ## What to do next Name the event before choosing the scheme. Fire, theft, a single dishonest finder and your own forgetfulness need different answers, and a scheme claiming to cover all four usually covers none of them well. Then pick the simplest scheme that covers the events you actually face. For most holders that is one offline seed in two distant locations with a tested restore. A passphrase belongs on top only if it is stored separately and documented somewhere that survives you. Finally, run the restore. A backup never restored is a belief. The test costs one withdrawal: $2 to $10 for an ERC-20 token on Ethereum L1 in FeeEdge’s 17 August 2026 sample, and a fraction of a cent on Base per Ethereum.org’s 1 to 6 October 2026 snapshot. The stablecoin float was $292.40B in October 2026, per this board’s data pack, and the difference between a tested backup and an untested one is the whole position. ### FAQ Q: Is a BIP39 passphrase the same as a second seed? A: No. A passphrase derives a different wallet from the same 24 words, so the seed alone opens an empty wallet. A second seed is an independent wallet with its own words. The first hides one wallet behind two secrets, the second creates two wallets that must each be backed up. Q: What happens if I forget the passphrase? A: The funds behind it are unrecoverable, and there is no error message to tell you the passphrase was wrong rather than the seed. Because the same 24 words open a valid but empty wallet, a typo looks like an empty account. Document that a passphrase exists, without documenting the passphrase in the same place. Q: Does splitting a seed make it safer? A: It removes the single object, which is real progress against a single finder. It also adds a reconstruction step that happens once, under stress, possibly years later. Shamir sharing with a threshold keeps availability; splitting halves of a word list does not. ## Restaking points: what you are actually underwriting URL: https://www.caijing188.com/restaking-points-risk/ Updated: 2026-10-27T22:09:58+08:00 Verified: 2026-10-27 ### Quick answer Restaking points are an unpriced claim on a token that may not exist, not a yield. The real term sheet is the slashing condition attached to each additional service, and the delegation bounds that decide whether your exposure can change without consent. Ethereum's exit queue held 850,000 ETH at about 14.8 days, as of October 2026. ### Key takeaways - A points balance has no price, no redemption right, and no conversion ratio until a token exists - Each additional service adds a slashing condition, and correlation runs through the operator - Unbounded delegation lets an operator add a service after you delegate, changing your risk without consent - October 2026 carried three exits: slashing exposure, vesting, and an 850,000 ETH queue at about 14.8 days ![A restaking interface listing points balances and operator delegations](https://img1.ladyww.cn/picture/Picture00437.jpg) Restaking sells a yield on top of a staking yield and pays part of it in points. Points are not a yield. They are an unpriced claim on a token that does not exist, issued by a team that has not published terms, against security work that can be slashed. What you are underwriting lives in the slashing conditions. ## Points are a claim, not a yield A point balance is a number in a database. It has no price, no redemption right, and no conversion ratio until a token exists and a distribution is announced. The honest model is an option: some probability of receiving some amount at some future date, with all three variables unknown. That is a description of what a points ledger is, not an accusation. The comparison is against the alternative use of the same ETH. Staking paid 2.63% gross in Q3 2026, within a 2.6% to 3.3% full-year range, and more than 96% of ETH staking rewards came from new issuance rather than fees per Bitwise’s Q4 2026 figures. A restaking layer adding points has added a claim on top of a yield that is already mostly new issuance ([staking yield is not income](/staking-yield-is-not-income/)). ## The slashing condition is the real term sheet Restaking lets staked ETH secure additional protocols, called actively validated services. Each brings rules, and each rule is a route by which principal can be reduced. The question is not how much a service pays. It is which action, by whom, causes a slashing event, and how much stake is exposed when it does. Some conditions are objective and enforced in code, such as signing two conflicting attestations. Others are committee judgements about liveness or operator behaviour, and those need reading, because a judgement call is a different risk from a cryptographic violation. Correlation is the part that gets missed. If one operator runs validators across several services and one service faults, slashing can reach stake that is backing everything else. Diversifying across services does not help when the operator is the shared component. The operator list matters more than the service list, and the question of [who secures a rollup](/who-secures-a-rollup/) has a name attached. ## Your exposure is the operator’s exposure only if delegation is bounded Delegation structure decides whether your risk is the risk you chose. A bounded delegation caps how much stake an operator can attract, restricts which services the operator may join, and requires the delegator to approve changes. An unbounded delegation lets the operator add a service after you delegate, so the terms you agreed to are not the terms you hold. Read for three things. Whether new services require your consent. Whether there is a cap per operator. And whether the penalty is proportional to the stake delegated to that operator or to the whole pool. The third is where the asymmetry hides: a small allocation to a high-risk service can expose the whole position if the penalty applies at pool level, turning a diversifying bet into a correlated one. ## The exit queue meets the points ledger Points usually vest, and the vesting clock starts on a date nobody outside the team controls. Meanwhile the underlying ETH has a queue of its own. Ethereum’s exit queue held 850,000 ETH in October 2026 at a wait of about 14.8 days, and total stake was 43.8 million ETH on 23 October 2026. A solo validator needs 32 ETH, so most restakers reach the queue through a protocol or a pooled wrapper, and the wrapper’s own liquidity is a separate constraint. The position therefore carries three locks: the service’s slashing exposure, the vesting schedule, and the validator exit queue. A user who models only the last has modelled the most visible lock and the shortest one. ## What the pitch leaves out The pitch is a number with a plus sign. The term sheet is a list of conditions under which the number turns negative. | What you are told | What you are underwriting | Where to check | |—|—|—| | Extra yield on your ETH | Additional slashing conditions, per service | Service documentation | | A points balance | An unpriced claim with unpublished terms | Programme terms, if any exist | | Operator performance | Correlation across that operator’s services | Operator and service lists | | Flexible exit | Queue, plus vesting, plus wrapper liquidity | Queue depth and vesting schedule | None of this is a reason to avoid restaking. It is a reason to size it as a venture position, and to keep the staked amount separate from anything you might need in a drawdown. There is a defensible version of the trade. Bounded delegation, a service list narrow enough to read, an operator with a limit you can name, and an amount you would not need to sell for a month. That is a position with a defined downside. ## Verdict Restaking points are not yield. They are an option on a future token, sold alongside a real and partly quantifiable increase in slashing exposure. Underwrite the slashing condition first, then the delegation bounds, then the exit. If the terms of a service can change after you delegate, you are not holding a yield. You are holding whatever the operator decides next. ### FAQ Q: Are restaking points a form of yield? A: No. Yield is a payment received in an asset with a price. Points are a ledger entry redeemable, if at all, at a future date on terms that have not been published. Modelling them as a rate imports a price and a date that do not exist yet. Q: What should be read before delegating to a restaking operator? A: The slashing conditions for each service the operator has joined, whether the penalty applies to the operator's delegated stake or the whole pool, and whether the operator can add services without your approval. Bounded delegation is the difference between a defined downside and an open one. Q: Does diversifying across operators reduce the risk? A: Only if the operators do not share services or infrastructure. Correlation in restaking usually enters through the operator, which may validate for several services at once. Spreading stake across operators that all run the same service set reproduces the same exposure. ## Why a clean wallet can still be frozen URL: https://www.caijing188.com/sanctions-screening-freeze/ Updated: 2026-10-27T22:09:58+08:00 Verified: 2026-10-27 ### Quick answer Screening scores the cluster of addresses around yours, so an inbound transfer from a counterparty two hops back can taint a wallet you never used badly, as of October 2026. Our board ranks 24 venues. Keep the trail and use a labelled withdrawal address, so an appeal has documents to work with. ### Key takeaways - Screening assigns risk to clusters of addresses, not to a single transaction - Inbound transfers carry the risk you did not choose, because the counterparty decided to send - A freeze is a venue action, so the appeal is a venue process with documents - Keep amount, asset, network, fee and address for every transfer, dated ![A wallet graph with a flagged address two hops from the centre](https://img1.ladyww.cn/picture/Picture00427.jpg) A wallet can be clean and still fail a screen. Screening does not read intent. It scores the addresses around yours, on a graph that includes counterparties you never chose, two and three hops back. Understanding that graph is the difference between a delay you can answer and a freeze you cannot. ## Screening scores a neighbourhood Analytics firms assign risk to addresses and to clusters. A cluster is a set of addresses believed to be controlled by one party, inferred from transaction patterns, deposit addresses and shared funding. The score is a property of the cluster and its exposure, not a verdict on one payment. That is why two customers with identical behaviour can get different results. They banked at different places, and the graph around each one looks different. ## Two hops is enough ### What the cluster includes The cluster typically includes addresses that funded yours, addresses your counterparties used, and services that touched the same flow. Shared deposit addresses at a venue can join unrelated customers into one graph node, which is one reason exchange withdrawals are scored as a group. The inference is probabilistic. It can be wrong, and it can be stubborn. ### Why inbound carries the risk An outbound transfer to a screened address is a decision you made. An inbound transfer is a decision someone else made about you. Venues see both, and the inbound one is harder to explain, because the counterparty chose to send and you may not know them. A deposit from a mixer, a sanctioned address, or a wallet flagged for fraud can move your cluster’s exposure even when the amount is small. Small amounts from bad places are a known pattern, which is why they are scored rather than ignored. ## The freeze is a venue action A public chain has no freeze function for an ordinary transfer. The block happens at the venue: custody of the balance, a hold on withdrawal, a request for documents. The venue acts because its banking and licensing relationships require it to, and because a supervisor can ask. So the appeal goes to the venue. The chain will confirm that a transfer happened. It will not review your case, and it has no one to review it. Token-level restrictions are the exception. Where an issuer can deny an address, the freeze can sit in the contract rather than at the venue, and the appeal runs to the issuer. That is a different process with a different timeline. ## What an appeal needs A dated record of the transaction, the counterparty if known, the source of the funds and the purpose. Venues vary in what they request, but the shape is consistent: they need to reconstruct a story from documents. Address labels help here. A labelled withdrawal address, used repeatedly, gives the venue a history that reads as one customer. A fresh address for every withdrawal gives them a set of unrelated transfers that happen to point at you. Consistency is also cheaper. A Solana SPL withdrawal runs about 0.5 to 1 USDT and an ERC-20 withdrawal $2 to $10 (FeeEdge, 17 August 2026), so batching to one known address is both a fee decision and a documentation one. ## Building a trail you can produce Keep the venue record of each transfer, not just the explorer link. Note the counterparty when it is a person. Keep the bank statement that funded the account. If a screen goes wrong, those three documents are the appeal. For the custody side, [how much a venue keeps in a hot wallet](/how-much-in-a-hot-wallet/) explains why a balance can be blocked at a venue while the chain shows it intact. For the venue’s own disclosure, [the three numbers in a reserve report](/proof-of-reserves-three-numbers/) is the companion read. | Signal | What it looks at | What you control | |—|—|—| | Direct counterparty | The address you transacted with | Who you accept funds from | | One-hop cluster | Addresses funded from the same source | How and where you receive | | Two-hop exposure | The graph behind your counterparty | Little, but you can document it | | Deposit pattern | Size and timing of inflows | Consistency over time | | Address reuse | The history attached to one address | Whether you label deliberately | ## What to do next Use one labelled address per purpose, and keep the receipt for every transfer. If a withdrawal is held, ask which list or score produced the hold and what document would clear it. That question is answerable, and it is the fastest route out. Our board ranks 24 venues as of October 2026, and the [exchange page](/exchange/) is where the structural comparison sits. Neither the board nor the chain can clear a freeze. Only the holder of the balance can, which is why the paperwork matters more than the transaction. A clean wallet is a starting condition, not a defence. The defence is a trail someone else can read. ### FAQ Q: How can my wallet be flagged if I never touched a sanctioned address? A: Because scoring works on a graph. Analytics firms group addresses into clusters using transaction patterns, deposit addresses and shared funding, then score the cluster's exposure. A counterparty two hops back can raise that exposure without you transacting with it directly. The score describes the neighbourhood, not your intent. Q: Can a blockchain freeze my funds? A: A normal transfer on a public chain cannot be reversed or frozen by the chain itself. The block happens where the balance is held: a venue restricting withdrawal, or an issuer adding an address to a token-level deny list. That distinction decides who you can appeal to, and the answer is usually the venue. Q: What documents help when a withdrawal is blocked? A: A dated record of the transfer, the counterparty if you know them, evidence of where the funds came from, and a short statement of purpose. Venues differ in what they ask for, but the shape is consistent: they need to reconstruct a story from documents rather than accept an explanation. ## Wallet drains start with a signature, not a hack URL: https://www.caijing188.com/drainers-start-with-a-signature/ Updated: 2026-10-27T22:09:59+08:00 Verified: 2026-10-27 ### Quick answer Most drains are a signature on a permit or a fake mint page, not a broken curve. Ethereum.org's L2 directory put Base at $14.6B of value as of October 2026, and an approval can move a balance without a key being stolen. Simulate, read the request, and keep savings behind a key that never connects. ### Key takeaways - A permit costs no gas now and can cost the whole balance later - Simulation models the same bytes you sign, so it cannot tell you the payload is honest - The spender address matters more than the amount on any approval - Savings behind a key that never connects to a site turns a drain into a small loss ![A wallet request dialog showing a permit signature request](https://img1.ladyww.cn/picture/Picture00457.jpg) A wallet drain is usually not a broken curve. The key stays where it should, and the funds leave because the owner signed a request that gave someone else permission to move them. The pattern repeats: a page that looks like a mint, a connect dialog, a signature that costs no gas, and a balance that disappears later. ## The common path The sequence has four steps and only one of them involves you. A link arrives promising an airdrop, a whitelist place, or a free mint. The page connects to your wallet and asks for a signature. The request is a permit, an approval or an operator approval, and it costs nothing to sign. Weeks or months later a transaction from an address you have never seen moves the assets. Nothing was hacked in the sense most people mean. No key was extracted, no firmware was exploited, no exchange was breached. A permission was granted, and permissions do not expire. ### Why the curve is not the target secp256k1 has not been broken, and attacking it costs more than asking you to approve something. The economics point at the consent layer, so the attacks sit there. So the defence is a procedure, not a chip: read the request, simulate it, refuse what you cannot understand. ## Simulate, but do not worship the simulation Transaction simulation runs the call against current chain state and shows the balance changes it expects. It is the best single pre-signature check available, and it has three limits. First, it models the same bytes you are about to sign. If the payload is hostile, the simulation shows the hostile outcome honestly, which is useful only if you read it. Second, state can change between simulation and execution. A contract can behave differently depending on a balance, a block number or a variable an attacker controls. Third, proxies and delegate calls can make the simulated effect differ from the label. A simulation that says transfer is not a guarantee about which contract executed. Treat a clean simulation as a filter, not a certificate. ## Read the request, not the page Four fields decide almost everything, and all four sit in the wallet popup rather than the website. What function is being called. A transfer moves a specific amount. An approval grants a permission. A permit grants a permission with no gas, which is the one people sign without noticing. Who receives the permission. For an approval this is the spender, and it matters most. A spender you cannot identify is a no. How much. Unlimited is a choice the interface makes for you, and it turns one mistake into a full balance. What it costs now. A zero fee request is not free. Its cost arrives later, when someone else uses the permission. ## Keep savings behind a different key The strongest structural control is separation. A wallet that connects to sites should hold roughly what you would carry in a physical wallet. Savings belong behind a key that has never seen a connect dialog and never will. That rule caps the damage of any signature mistake at the spending balance, and it makes the decision easier, because the answer no longer involves your savings. Address poisoning is the companion risk. A lookalike matching the first and last characters turns a copy and paste into a transfer to a stranger. The habit that closes it is in [the address poisoning note](/address-poisoning-habit/). | Request | What it can move | Revocable later | Field to check | |—|—|—|—| | Token transfer | A stated amount | No | Full destination address | | Finite approval | Up to the allowance | Yes | Spender and amount | | Unlimited approval | The whole balance | Yes, by revoking | The word unlimited | | Permit signature | The whole balance | No, once used | Why it needs no gas | | Operator approval | Every token in a collection | Yes, by revoking | The operator address | | Batch call | Several of the above | Depends on the calls | Each action in the batch | ## Where account abstraction changes the answer Smart accounts can add session keys, spending limits and allowlists, which shrinks the blast radius of a bad signature. They do not remove the need to read the request, and they add a surface: the session key can be the thing that drains the account. That trade is set out in [the limits of account abstraction](/account-abstraction-limits/). The short version is that a policy is only as good as the code enforcing it, and that code is another contract you are trusting. ## What to do next Simulate every request you do not recognise, and refuse the ones you cannot decode. Enable the setting that blocks blind signing on the device that holds savings. Then separate the keys. Move savings behind a wallet that never connects to a site, and keep a small balance for experiments. The [hot wallet sizing question](/how-much-in-a-hot-wallet/) covers how much that small balance should be. Finally, size the mistake. Ethereum.org’s L2 directory put Base at $14.6B of value in October 2026, and all of it moves on signatures. The loss is set by the balance behind the key, not by the fee in front of it. ### FAQ Q: Can a drainer take funds without my private key? A: Yes, through a permission you signed. A permit, an approval or an operator approval lets a contract move assets later, and the signature costs nothing at the time. The key stays in the device and the balance still leaves. Q: Does simulation stop drainers? A: It catches most of them, because a hostile payload simulated honestly shows the hostile outcome. It cannot certify intent: state can change before execution, and proxies can make the simulated effect differ from the label. Treat a clean simulation as a filter, not a guarantee. Q: What should I check before signing anything unfamiliar? A: Four fields: the function being called, the spender receiving the permission, the amount with unlimited treated as a red flag, and whether the request costs no gas. If any of the four is unclear, refuse and open the site again from a bookmark rather than the link you were sent. ## Stablecoin supply hit $292B, and the holders are narrower than the number suggests URL: https://www.caijing188.com/stablecoin-supply-who-holds-it/ Updated: 2026-10-27T22:09:59+08:00 Verified: 2026-10-27 ### Quick answer Stablecoin supply stood at $292.40B in October 2026, but the float that actually trades is concentrated in a handful of issuers and a handful of chains. A large supply number is not the same as deep liquidity. ### Key takeaways - Watch issuer share, not just total supply. - Check which chain holds the float you plan to use. - Trading depth matters more than the headline. ## A stock, not a flow Stablecoin supply is a stock figure: it counts tokens that exist at a moment in time. The number that decides whether your order fills is the flow — how much of that stock actually changes hands on the venue you are using. ## Where the float sits Three issuers dominate, and the distribution of holders is narrower than the headline suggests. When a large share of supply belongs to trading desks and treasuries, it behaves like inventory rather than like circulating money. | Measure | What it tells you | What it hides | |—|—|—| | Total supply | Size of the stock | Whether it trades | | Holder count | Breadth of ownership | Size of each holder | | On-chain transfer volume | Real usage | Who is moving it | ## The practical test If you are moving size, check the depth on the exact chain and venue you plan to use, then check the redemption path back to fiat. The supply number will not answer either question. ### FAQ Q: Does a bigger stablecoin supply mean better liquidity? A: No. Supply counts tokens that exist; liquidity counts tokens that trade. A large float held by a few treasuries does little for a $50,000 order. Q: Why does the chain matter? A: Because your exit route is chain-specific. A float concentrated on one network is thin everywhere else, and a bridge adds a second risk.