Public — correction log
Every published editorial mistake
Factual errors, misapplied scoring, stale data and claims the site could not back. Each entry says what was wrong, what was fixed, how it was caught and whether the published order moved. An entry whose logged date is later than its corrected date was copied here from the methodology or data changelog that first published the fix. No entry is ever deleted.
To report something wrong on the site, open the report form: it needs no mail app and no email address, gives you a reference to look the report up, and falls back to email (mathias@cryptopaymentfinder.com) if it cannot reach us. Reviewed within 7 days. Every provider page also carries a report link on each criterion row, with the value it printed filled in.
No reader report has been logged yet. When a reader’s report leads to a correction, the entry says so and names the report. The reader is named only if they asked to be; everyone else is counted, never named. No contact detail is ever published.
No reader has been credited by name yet.
No provider has replied yet. A provider may reply to what we publish about it. The reply is printed beside our verdict on its provider pages, under its name, with the date we received it and the date of the verdict it answers, edited only to remove contact details and links. A reply changes no score, rank or verdict: the code that orders a ranking cannot read these records, and the build fails if it can.
No provider has been found to have given us false information. If we find that a provider gave us false or materially misleading information, its provider pages carry a dated note saying so for 12 months from the finding, and it may not submit replies or evidence until a review restores that right. The finding is logged as a correction, which the note names. The note itself changes no score; what was false is removed and re-read from the provider's own published sources.
How disputes are answered, by quarter
Every reader report and provider submission is a ticket with who sent it, why, the criterion and rung in question, and the days it arrived, was first answered and was decided. Our promises: a person answers within 2 business days, a report is reviewed within 7 days, nine in ten are decided within 14 days, a reader-safety matter is answered within 24 hours, and we aim to keep each promise at least 95% of the time. Appeals are not counted yet: there is no appeal route.
No dispute has been logged yet, so there is nothing to measure. The current quarter is still published below, labelled, so a quiet quarter cannot be told apart from a missing one.
| Quarter | Received | By | Answered on time | Decided in (median, p90) | Review promise kept | Safety answered in 24 h | Accepted | Changed a score |
|---|---|---|---|---|---|---|---|---|
| 2026-Q4 | 0 (0 open) | No dispute received this quarter yet, so nothing is measured. | none yet | none yet, none yet | none yet | none | none yet | none yet |
disputes.For providers: reply, send evidence, ask for a correction, claim your listing
Any provider we list can reply to our verdict, send evidence, ask for a correction or claim its listing. Each message gets a public ticket number and is listed here with its status. None of it is read by the code that ranks: a ranking moves only when we re-read a public page, or enter a signed statement under the attestation rule.
- Reply to our verdict: Your answer to what we publish about your product, up to 1,200 characters. Printed beside our verdict on your provider pages, under your name, with the date we received it. It changes no score, rank or verdict.
- Send evidence for a criterion: A public page, a quote or, where no public page can carry the fact, a signed statement. We re-read the page against the published ladder. A signed statement is labelled attested-by-provider and lifts a criterion by at most one band over what public evidence supports.
- Ask for a correction: A fact we publish that is wrong: a fee, a licence, a supported country, a date. Checked against your own published source; a fact we change is logged as a correction on this page.
- Claim your listing: Name the person we write to about your product. Accepted once a message from an address at your own domain, or a DNS TXT record we send you, proves it. Claiming a listing changes nothing about how it is ranked.
No provider has written yet: the intake opened on 2026-10-15. The first ticket will be listed here with its number, its status and what it became.
Gemini e
Gemini e
The OTC desks review of 10 November 2026, an objection to Gemini’s footing score, and an independent model review run on the first place.
Yes. Gemini e
Kraken OTC Desk, second in OTC desks at 77, was ranked on pages whose own terms bar automated reading: Kraken’s general terms (section 9) bar “any web scraping, web harvesting, or data extraction methods” and “any bots, robots, parsers, spiders, scripts” on its services.
Kraken OTC Desk leaves the ranking until a person reads its pages, listed under “Not confirmed”, and its earlier scores are not shown. None of it was found wrong.
A review of every OTC desk’s own terms on 11 November 2026.
Yes. Kraken OTC Desk, second at 77, leaves the order; B2C2 moves from third to second (CC-2026-0171 gives every score). In the country lists of Germany, France, Spain, Italy, the Netherlands, Portugal, Ireland, Austria, Belgium, Poland, Sweden, Finland, Norway, the United Kingdom and Australia, where Kraken OTC Desk was first, B2C2 is now first; in Switzerland, Hong Kong, South Korea, Thailand, Indonesia, the Philippines, Vietnam, the United Arab Emirates, the United States, Canada, Japan and Singapore Gemini e
Symbiosis was ranked fifteenth in bridges at 41 on scores read before its native Bitcoin bridge was exploited on 11 September 2026. The attacker minted unbacked tokens from that bridge and took about $336,000; Symbiosis paused the bridge and later recovered 15 BTC (Cointelegraph, 14 September 2026, citing the security firm Blockaid).
Symbiosis is held: no Symbiosis page read on 10 November 2026 says whether its Bitcoin bridge has reopened, so the chains it can move funds between could not be confirmed. It is listed under “Not confirmed” with the exploit, and its earlier scores are not shown until it can be read again.
The bridges review of 10 November 2026 and an independent second reader’s pass the same evening.
Yes. Symbiosis, fifteenth at 41, leaves the order (CC-2026-0166 gives the order that remains).
Hop Protocol stayed in the bridges list after it was withdrawn on 23 October 2026, last at 41.
Hop Protocol is retired from the bridges list. On 10 November 2026 hop.exchange served an unrelated online gambling page, and its app and its route service answered only with errors, as on 30 September and 23 October 2026. Its domain is not linked from this site.
The bridges review of 10 November 2026.
Yes, at the bottom only. Hop Protocol, seventeenth at 41, leaves the order; it was already shown as withdrawn.
Jumper Exchange and Squid Router were shown as charging no bridge fee. Squid’s fee page now states a convenience fee on some front-end routes and no rate; LI.FI’s fee FAQ states a 0.25% service fee, which its quote API charges the default integrator but not Jumper’s routes (quoted the same day at 0 to 0.02%).
Neither publishes a rate, so neither is shown with a fee of zero any more; both have also left the ranking until their pages can be read (CC-2026-0166). Orbiter Finance moves from sixth to second, 61 → 64, on its chain coverage (eleven VM families over 74 mainnets, no Bitcoin family); no provider changed a price.
The bridges review of 10 November 2026: the providers’ own fee pages and quotes on the modelled route, read three times that day, and an independent model review run on the top three.
Yes. Orbiter Finance moves from sixth to second, 61 → 64; Jumper Exchange (second, 63) and Squid Router (third, 62) leave the order (CC-2026-0166).
Eight bridges were ranked on pages we can no longer read the way this site reads a provider. The terms of Wormhole, Squid Router, Stargate and the Scroll bridge bar reading them with robots, crawlers or scrapers; on 10 November 2026 Jumper’s route comparison opened only after an amount is typed, Bungee’s terms address led to its home page, Synapse’s quote service found no route and Everclear’s site did not open.
Jumper Exchange, Squid Router, Bungee, Scroll Bridge, Synapse, Stargate, Wormhole and Everclear (formerly Connext) leave the ranking until they can be read, each listed under “Not confirmed” with its reason, and their earlier scores are not shown. None of them left because of anything we found wrong about it. Optimism Gateway, read again on 10 November 2026 from its own documentation, now scores 2 for speed (was 3) for its seven-day withdrawal and 9 for trust (was 10): 55 → 52.
A review of every bridge’s own terms and pages on 10 and 11 November 2026, checked by an independent second reader.
Yes. Relay stays first at 86. Orbiter Finance moves from sixth to second, 61 → 64 (CC-2026-0167); Base Bridge moves from fifth to third at 61; Arbitrum Bridge moves from seventh to fourth at 55; Optimism Gateway moves from eighth to fifth, 55 → 52; zkSync Bridge moves from twelfth to sixth at 50; deBridge moves from fourteenth to seventh at 46. Jumper Exchange (second, 63), Squid Router (third, 62), Bungee (fourth, 62), Scroll Bridge (ninth, 53), Synapse (tenth, 51), Stargate (eleventh, 51), Wormhole (thirteenth, 50) and Everclear (formerly Connext) (sixteenth, 41) leave the order, as do Symbiosis (CC-2026-0169) and Hop Protocol (CC-2026-0168). Each of the 27 country lists (Germany, France, Spain, Italy, the Netherlands, Portugal, Ireland, Austria, Belgium, Poland, Sweden, Finland, the United Kingdom, Switzerland, Norway, the United States, Canada, Australia, Japan, Singapore, Hong Kong, South Korea, Thailand, Indonesia, the Philippines, Vietnam and the United Arab Emirates) now reads Relay, Orbiter Finance, Base Bridge, Arbitrum Bridge, Optimism Gateway, zkSync Bridge and deBridge, in that order.
Bitrefill, Coincards, CoinGate Gift Cards, SpendCrypto and BitGift are no longer ranked in the crypto gift card list, and with them gone the list has too few products to rank: we could read three of its eight in full, and a ranking needs five. None of them left because of anything we found wrong about it.
On 11 November 2026 the terms of Bitrefill and SpendCrypto bar automated reading, BitGift’s robots.txt asks AI agents not to read its pages, and the pages of Coincards and CoinGate turned our reader away. We follow what a site asks, so we could not read their prices and terms the way this site reads a provider. Their earlier results were not found wrong. The list shows Bidali, Coinsbee and CryptoRefills A to Z with what we read, with no rank, until five can be read: a person reads two more by hand, or a provider allows our reader.
A review of every gift card store’s own terms and robots.txt on 10 and 11 November 2026, which left fewer products than a ranking needs.
Yes. Crypto gift cards publish no order while they are not ranked; until 12 November 2026 they had Bitrefill first (81), Coincards second (72), CryptoRefills third (71), Coinsbee fourth (62), CoinGate Gift Cards fifth (61), SpendCrypto sixth (53), Bidali seventh (51) and BitGift eighth (41). In the 27 country lists (Germany, France, Spain, Italy, the Netherlands, Portugal, Ireland, Austria, Belgium, Poland, Sweden, Finland, the United Kingdom, Switzerland, Norway, the United States, Canada, Australia, Japan, Singapore, Hong Kong, South Korea, Thailand, Indonesia, the Philippines, Vietnam and the United Arab Emirates) Bitrefill was first in each, and each list shows nothing while the category is not ranked.
Koinly, Blockpit, CoinTracker, ZenLedger, Waltio, Ledgible and Summ (formerly Crypto Tax Calculator) are no longer ranked in the crypto tax software list, and with them gone the list has too few products to rank: we could read four of its eleven in full, and a ranking needs five. None of them left because of anything we found wrong about it.
Their own terms bar reading their pages with bots, crawlers or other automated means, and Koinly’s and CoinTracker’s pages also turned our reader away with a block page; Summ’s help centre asks readers like ours, in its robots.txt, not to read its pages. We follow what a site asks, so we could not read their prices and terms the way this site reads a provider. Their earlier results were not found wrong. The category is listed without an order until five can be read: a person reads one more by hand, or a provider allows our reader.
A review of every tax software provider’s own terms and robots.txt on 10 and 11 November 2026, which left fewer products than a ranking needs.
Yes. Crypto tax software publishes no order while it is not ranked; until 12 November 2026 it had Koinly first (81), Blockpit second (80), CoinLedger third (78), Divly fourth (78), Summ (formerly Crypto Tax Calculator) fifth (77), CoinTracker sixth (76), Waltio seventh (75), ZenLedger eighth (73), CoinTracking.info ninth (72), TokenTax tenth (66) and Ledgible eleventh (62). In the 27 country lists (Germany, France, Spain, Italy, the Netherlands, Portugal, Ireland, Austria, Belgium, Poland, Sweden, Finland, the United Kingdom, Switzerland, Norway, the United States, Canada, Australia, Japan, Singapore, Hong Kong, South Korea, Thailand, Indonesia, the Philippines, Vietnam and the United Arab Emirates) Koinly was first in each, and each list shows nothing while the category is not ranked.
Idle Yield Tranches (Senior AA and Junior BB), second in structured products, said in the present tense that the Junior tranche on its largest pool "holds about $21 against about $297,000 of Senior (0.0072%)", a figure read on 28 September 2026, and did not say that Idle’s own homepage now leads with "Idle is now Pareto – the credit coordination protocol". On 10 November 2026 that pool held 78,692.82 USDC, with Junior worth about $41 against about $78,641 of Senior (0.052%).
The row dates its stale $21/$297,000 sentence and publishes the 10 November contract read, with the pool’s contract address and block, and the rebrand as a watch-out. Idle keeps its "Counterparty backing the payoff" score of 6 and second place because that criterion’s mid band is "a disclosed first-loss tranche" and the same near-zero cover that set 6 on 28 September (0.0072%, now 0.052%) is present today, while pool shrinkage and the Pareto rebrand are not inputs to any structured products criterion.
An incident follow-up on 10 November 2026: Idle’s homepage and the Yield Tranche contracts read on-chain, checked by an independent second reader the same evening.
No. No score and no place changed: Idle Yield Tranches stays second in structured products at 70, behind SOFA.org (Earn and Surge) at 82 and ahead of Ithaca Protocol at 69.
Pirex (px
The sponsor’s sunset notice, which does not name px
An incident follow-up on 10 November 2026: Dinero’s homepage, its docs and forum hosts, and the PirexCvx contract read on-chain, checked by an independent second reader the same evening.
No. No score and no place changed: the order Convex Finance (cvx
MetaMask Transaction Shield's record did not mention the infrastructure incident MetaMask disclosed, identified on 28 September 2026, in which "approximately 0.385 ETH in rewards was diverted" in MetaMask Staking; the MetaMask wallet row already carries it in its incident record.
Transaction simulation tools are not ranked since 11 November 2026. Transaction Shield is listed by name on that page and on its own row page, with why it was not read this time, but its record is not shown, so no watch-out is printed for it; the incident is already in the MetaMask wallet row's incident record, which is unchanged. MetaMask's notice (metamask.com/news/user-update, updated 5 November 2026) says the incident did not involve Transaction Shield.
An incident follow-up on 10 November 2026: MetaMask’s own notice, read again and checked by an independent second reader the same evening.
No. No score and no place changed: transaction simulation tools are not ranked, and the order of crypto wallets is unchanged.
Hexagate (by Chainalysis), third in monitoring platforms, carried no line about Chainalysis’s October–November 2026 security incident: on 4 November Chainalysis’s status page said users "do not have access to the Hexagate frontend, the API, and GateSigner" while monitoring and alert delivery kept running, and all customer API keys were invalidated.
The row publishes the dated incident as a watch-out, quoting the status page as read on 10 November 2026, and its review line names the read. No monitoring platforms criterion names a vendor breach or an outage, so no score and no position moves.
An incident follow-up on 10 November 2026: status.chainalysis.com and Chainalysis’s incident post, checked by an independent second reader the same evening.
No. No score and no place changed: Hexagate (by Chainalysis) stays third in monitoring platforms at 72, behind Hypernative (79) and BlockSec Phalcon (78) and ahead of Forta Firewall (68).
Chainalysis, second in wallet screening tools, carried no line about the security incident it disclosed on 30 October 2026, in which, by its update of 4 November, "names, usernames, password hashes and live tokens for all Chainalysis users were exposed to a malicious actor", and KYT and Address Screening were offline from 4 to 7 November. Our 3 November re-read recorded the incident for the watch-outs, but the step that applies a re-read to a row writes no incidents, so it never reached the page.
The row publishes the dated incident as a watch-out, with "as applicable" kept in the password-reset sentence, and its review line names it. No wallet screening criterion names a vendor breach or an outage, so no score and no position moves; Chainalysis states it has "no evidence that any investigations data has been accessed or Chainalysis attribution data has been manipulated".
An incident follow-up on 10 November 2026: Chainalysis’s incident post and status page, checked by an independent second reader the same evening.
No. No score and no place changed: Chainalysis stays second in wallet screening tools at 82, behind TRM Labs (83) and ahead of Elliptic (78).
t
The row now publishes the 30/31 October to 9 November pause and the Aave freeze (still frozen at Ethereum block 26387780, 23:24 UTC on 10 November 2026) with the honest caveat that we could not read Threshold’s own notices or the pause mechanism, and the record’s "incidents none_found" is replaced with the dated incident. t
An incident follow-up on 10 November 2026: LlamaRisk’s post on Aave’s governance forum and the Aave V3 reserve read on-chain, checked by an independent second reader the same evening.
No. No score and no place changed: t
CC-2026-0116 left out one move: Zeal also moved from eighth to seventh in crypto cards. Its list of the rows that move up as rows leave named Coinbase Card (seventh to sixth) and Plutus Card (ninth to eighth) only.
This entry adds the move. CC-2026-0116 stays as it was published. Measured on rankings.json against the base CC-2026-0116 names (df6baed55): RedotPay Card, sixth, left the list, and Coinbase Card, Zeal and Plutus Card each moved up one.
An internal audit of the corrections log (2026-11-11) checked every move each recent entry names against rankings.json and data/country-orders.json.
No. No published score or rank changed; this names a move CC-2026-0116 measured and did not list.
Five rows whose exit runs through Binance withdrawals carried no warning after Binance's notices from 30 October 2026 described unauthorised withdrawals, and its status page still showed partial withdrawal suspensions for BTC, ETH, USDT and BNB on 10 November 2026: BTCB, WBETH, Binance P2P, Binance OTC and Binance Launchpool.
Each now opens its watch-outs with the warning its Binance exchange row carries, dated 11 November 2026: redeeming BTCB or WBETH, or withdrawing funds held on the other three, may be delayed or unavailable until the suspensions are lifted.
A reader-harm sweep of the last 45 days (HARM-4), checked the same day against the Binance exchange record and the rekt.news account.
No published score or place changed.
Balancer was ranked as an operating DEX and its V3 and V2 vaults as operating flash-loan venues after BAL holders voted to wind it down: its pools began moving to withdrawals-only on 30 October 2026, and bug-bounty cover ended that day.
The three rows carry a sunset date of 30 October 2026 and open with a dated warning: Balancer is shutting down, so do not add liquidity or build a new flash-loan integration on it, and withdraw a Balancer LP position. No page draws a Visit for them.
A reader-harm sweep of the last 45 days (HARM-4), verified the same day against the rekt.news account of 14 October 2026.
Yes. In decentralised exchanges, Balancer falls from tenth at 43 to eleventh, last, at 41, and Raydium moves from eleventh to tenth, its score unchanged at 41. In flash loans, Balancer V3 falls from seventh at 61 to eighth at 41, and Balancer V2 from eighth at 61 to seventh at 41; both now sit at the floor. Balancer is removed from the decentralised exchange list, and Balancer V2 and Balancer V3 from the flash loan list, on the country pages for Australia, Austria, Belgium, Canada, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, the United States and Vietnam, which list only products that operate.
Mercado Bitcoin's held row said only that its pages could not be read, while its hot wallets and customer deposit addresses had been drained on 2 November 2026 and withdrawals and crypto transfers were still closed on 10 November.
Its "Not confirmed" line now reads "Re-read pending, page walled: drained on 2 November 2026; on 10 November withdrawals and crypto transfers were still closed", and its exclusion text and the tokenised-bonds refusal open with a stopped warning: do not send crypto to a Mercado Bitcoin deposit address until transfers are back.
A reader-harm sweep of the last 45 days (HARM-4), verified the same day against the rekt.news account of 10 November 2026.
No published score or place changed: Mercado Bitcoin has not been ranked since 10 November 2026.
Electrum's record carried no warning after Electrum's security advisory of 24 October 2026 said that in versions before 4.8.3 a malicious server can trick a user into paying an attacker through a fee bump, and that the attack is happening in the wild.
Electrum's record now opens its watch-outs with a warning dated 11 November 2026: update to Electrum 4.8.3 or later now, and until then verify the output addresses and amounts of every transaction before signing, fee bumps included.
A reader-harm sweep of the last 45 days (HARM-4), verified the same day against Electrum's own security advisory.
No published score or place changed.
Off-ramps were ranked across seven products. On 11 November 2026 four of them were held as not confirmed: Kraken, ranked first, Ramp Network, ranked second, and MoonPay, ranked sixth, because their own terms bar reading their pages with automated means (CC-2026-0137, CC-2026-0136, CC-2026-0135), and Coinbase, ranked third, because its terms page answered only behind a bot check and has not been read (CC-2026-0146). We could read three of the seven in full, and a ranking here needs five.
Off-ramps are not ranked right now. The page lists Banxa, Bitpanda and Transak alphabetically, each read in full on 22 October 2026, with no rank, no first place and no scores, and it says what left and why. Its country pages forward to that page. The category is ranked again when five off-ramps can be read. If a person's read of Coinbase's user agreement finds no clause that bars automated reading, Coinbase's hold is reverted in a later correction, which leaves four off-ramps: still one short of a ranking.
The same review as CC-2026-0137, which left fewer off-ramps than a ranking needs.
Yes. Off-ramps publish no order while they are not ranked; until 11 November 2026 they had Kraken first (82), Ramp Network second (76), Coinbase third (58), Transak fourth (58), Banxa fifth (53), MoonPay sixth (53) and Bitpanda seventh (44). In the 27 country lists (Australia, Austria, Belgium, Canada, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, the United States and Vietnam) Kraken was first in 26 and Ramp Network in Japan, and each list shows nothing while the category is not ranked.
Crypto cards were ranked across eight products. On 11 November 2026 four of them were held as not confirmed: ether.fi Cash, ranked first, MetaMask Card, ranked third, and Nexo Card, ranked fourth, because their own terms bar reading their pages with automated means (CC-2026-0140, CC-2026-0138, CC-2026-0139), and Coinbase Card, ranked sixth, because Coinbase's terms page answered only behind a bot check and has not been read (CC-2026-0145). RedotPay Card has been held since 10 November 2026 (CC-2026-0116). We could read four of the nine in full, and a ranking here needs five.
Crypto cards are not ranked right now. The page lists Bleap Card, Crypto.com Card, Plutus Card and Zeal alphabetically, each read in full on 10 November 2026, with no rank, no first place and no scores; Zeal keeps its dated line that its app is winding down. The page says what left and why, and its country pages forward to it. The category is ranked again when five crypto cards can be read. If a person's read of Coinbase's user agreement finds no clause that bars automated reading, Coinbase Card's hold is reverted in a later correction, five cards are read and the category is ranked again.
The same review as CC-2026-0140, which left fewer crypto cards than a ranking needs.
Yes. Crypto cards publish no order while they are not ranked; until 11 November 2026 they had ether.fi Cash first (79), Crypto.com Card second (77), MetaMask Card third (69), Nexo Card fourth (68), Bleap Card fifth (60), Coinbase Card sixth (52), Zeal seventh (43) and Plutus Card eighth (41). In the 27 country lists (Australia, Austria, Belgium, Canada, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, the United States and Vietnam) ether.fi Cash was first in 26 and Crypto.com Card in Canada, and each list shows nothing while the category is not ranked.
Transaction simulation was ranked across nine products. On 11 November 2026 five of them were held as not confirmed because their own terms bar reading their pages with automated means: MetaMask Transaction Shield, ranked first (CC-2026-0134), and Blockaid, Rabby Wallet, Phantom and Webacy, ranked second, third, fourth and ninth (CC-2026-0144). We could read four of the nine in full, and a ranking here needs five.
Transaction simulation is not ranked right now. The page lists GoPlus Security, Kerberus (Pocket Universe), Scam Sniffer and Tenderly alphabetically, each read in full on 22 October 2026, with no rank, no first place and no scores, and it says what left and why. Its country pages forward to that page. The category is ranked again when five products can be read.
The same review as CC-2026-0144, which left fewer products than a ranking needs.
Yes. Transaction simulation publishes no order while it is not ranked; until 11 November 2026 it had MetaMask Transaction Shield first (81), Blockaid second (69), Rabby Wallet third (69), Phantom fourth (66), GoPlus Security fifth (65), Kerberus (Pocket Universe) sixth (62), Tenderly seventh (58), Scam Sniffer eighth (58) and Webacy ninth (51). In the 27 country lists (Australia, Austria, Belgium, Canada, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, the United States and Vietnam) MetaMask Transaction Shield was first in 26 and Blockaid in the United Kingdom, and each list shows nothing while the category is not ranked.
ether.fi Liquid was ranked sixth in yield vaults on ether.fi's help-centre articles and documentation, read by our research agent on 3 November 2026. ether.fi's terms of use say "you must not use any robots, spiders, crawlers, scrapers, or other automated tools or interfaces that are not provided by us to access the Website and extract data", and its help centre and documentation are part of that website, so a read made that way cannot date the row. Each of its four scores is a judgement of what ether.fi discloses, and no other primary source we may read states it.
ether.fi Liquid leaves the ranking and is listed under "Not confirmed": "Re-read pending, its terms bar automated reads: ether.fi's terms bar robots, crawlers and scrapers on its website." Its earlier scores are not shown. It returns when a person reads its pages in a browser.
A permitted-source check of the rows read from sites whose terms bar automated reading: a row stays only if every scored item comes from a source we may read, and none of this row's did.
Yes, in yield vaults. Measured on rankings.json and data/country-orders.json against the published rankings before this change; no score moves. ether.fi Liquid (#6, 62) leaves; Upshift moves #7 58 → #6 58, and the five rows above keep their places (Yearn first). ether.fi Liquid leaves the yield-vault lists of the 24 countries that listed it (Australia, Austria, Belgium, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates and Vietnam), where Upshift moves up one; no country's first place changes.
Kraken ETH Staking was ranked second in retail staking on Kraken's staking page and its staking support article, read by our research agent on 28 October 2026. Kraken's general terms bar "any web scraping, web harvesting, or data extraction methods" and "any bots, robots, parsers, spiders, scripts" on its sites, so a read made that way cannot date the row. Its commission, its exit wait and the rate it shows are published only on those pages; Kraken's terms documents refer to them without stating them.
Kraken ETH Staking leaves the ranking and is listed under "Not confirmed": "Re-read pending, its terms bar automated reads: Kraken's terms bar bots, scripts and scraping on its pages." Its earlier scores are not shown. It returns when a person reads its pages in a browser.
A permitted-source check of the rows read from sites whose terms bar automated reading: a row stays only if every scored item comes from a source we may read, and none of this row's did.
Yes, in retail staking. Measured on rankings.json and data/country-orders.json against the published rankings before this change; no score moves. Kraken ETH Staking (#2, 56) leaves; Lido (st
Kraken Wrapped Bitcoin (k
Kraken Wrapped Bitcoin (k
A permitted-source check of the rows read from sites whose terms bar automated reading: a row stays only if every scored item comes from a source we may read; two of this row's four could be read from Kraken's terms documents, and two could not.
Yes, in wrapped assets. Measured on rankings.json and data/country-orders.json against the published rankings before this change; no score moves. Kraken Wrapped Bitcoin (k
Coinbase was ranked third in off-ramps on its help pages on selling and withdrawals and its licences page, read by our research agent in October 2026. Coinbase's user agreement, which says how its sites may be used, answered our requests only behind a bot check, which we never bypass, so we have not read whether its terms allow automated reading. A row on terms we have not read cannot be dated.
Coinbase leaves the off-ramp ranking and is listed under "Not confirmed": "Re-read pending, page walled: Coinbase's terms page answers only behind a bot check, so we have not read it." Its earlier scores are not shown. If a person's read of Coinbase's user agreement finds no clause that bars automated reading, this hold is reverted in a later correction and the row returns on its reads.
A review of every ranked row against the terms of the sites it was read from; Coinbase's terms could not be read.
Yes, see CC-2026-0152: off-ramps are not ranked right now.
Coinbase Card was ranked sixth in crypto cards on its card terms and Coinbase's status page, read by our research agent on 10 November 2026. Coinbase's user agreement, which says how its sites may be used, answered our requests only behind a bot check, which we never bypass, so we have not read whether its terms allow automated reading. A row on terms we have not read cannot be dated.
Coinbase Card leaves the ranking and is listed under "Not confirmed": "Re-read pending, page walled: Coinbase's terms page answers only behind a bot check, so we have not read it." Its earlier scores are not shown. If a person's read of Coinbase's user agreement finds no clause that bars automated reading, this hold is reverted in a later correction and the row returns on its reads.
A review of every ranked row against the terms of the sites it was read from; Coinbase's terms could not be read.
Yes, see CC-2026-0151: crypto cards are not ranked right now.
Blockaid, Rabby Wallet, Phantom and Webacy were ranked second, third, fourth and ninth in transaction simulation on their own product, help and documentation pages, read by our research agent on 22 October 2026. Each one's own terms bar reading those pages with automated means: Blockaid's bar any automated system, "including without limitation, 'robots' and 'spiders'"; Rabby's bar scraping "except as permitted by a Paid Plan"; Phantom's bar scraping "from the App or Site"; and Webacy's bar navigating its service "with any tool, software, agent, engine, or other means other than through Webacy's provided interfaces". A read made that way cannot date a row, and none of the four rows had a scored item from a source we may read.
The four leave the ranking and are listed under "Not confirmed", each with a reader line that its terms bar automated reads, and their earlier scores are not shown. Each returns when a person reads its pages in a browser.
A second review of the rows next in line for first place after MetaMask Transaction Shield was held (CC-2026-0134), against each row's own terms, before any new order was published.
Yes, see CC-2026-0150: transaction simulation is not ranked right now. No new first place was published.
Aave GHO's pricing and verification notes quoted Aave's documentation on how GHO's rate may change: "no more than 500bps per 2-day period, up to a maximum 25% APR". aave.com's terms allow automated access only where it is expressly permitted, so that page cannot be cited, and a read of the contracts on chain found a one-day delay on the rate steward and no 25% cap.
The sentence leaves both notes; they keep the on-chain rate reads of 22 October 2026. No new claim on the rate control is made until a second reader confirms it on chain.
A review of the rows read from sites whose terms bar automated reading, checked against the contracts on chain.
No. No published score or rank changed.
MoonPay was ranked third in on-ramps on its pricing disclosure, terms, licences and help pages, read by our research agent on 10 November 2026. MoonPay's terms of use say "you will not use an anonymizing proxy, or any other automatic devices, spider or manual processes to copy or monitor our websites without our prior written permission", so a read made that way cannot date the row.
MoonPay leaves the on-ramp ranking and is listed under "Not confirmed": "Re-read pending, its terms bar automated reads: MoonPay's terms bar spiders and automatic devices that copy its websites." Its earlier scores are not shown. It returns when a person reads its pages in a browser.
A review of the other rows read from MoonPay's and Ramp's sites, after their off-ramp rows were held for the same terms.
Yes, in on-ramps. Measured on rankings.json and data/country-orders.json against the published rankings before this change; no score moves. MoonPay (#3, 70) leaves, with Ramp Network (#4, 70; CC-2026-0141). Transak stays first and Coinify second. Banxa moves #5 68 → #3 68, Coinbase Onramp #6 60 → #4 60, Mercuryo #7 57 → #5 57, Topper (Uphold) #8 54 → #6 54, Stripe Crypto Onramp #9 54 → #7 54, Onramp.money #10 48 → #8 48, Alchemy Pay #11 41 → #9 41, Koywe #12 41 → #10 41, Onmeta #13 41 → #11 41 and Fonbnk #14 41 → #12 41. MoonPay and Ramp Network leave the on-ramp lists of all 27 countries (Australia, Austria, Belgium, Canada, Finland, France, Germany, Hong Kong, Indonesia, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, Poland, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, the United States and Vietnam), and every row below them moves up; in the United Arab Emirates Banxa moves to first, where MoonPay was.
Ramp Network was ranked fourth in on-ramps on its fee and verification help pages, its terms and its licence page, read by our research agent on 10 November 2026. Ramp's terms of service, which count its website among its services, say users won't "reverse engineer, copy, modify, data scrape, automate or re-sell any part of the Services", so a read made that way cannot date the row.
Ramp Network leaves the on-ramp ranking and is listed under "Not confirmed": "Re-read pending, its terms bar automated reads: Ramp's terms bar data scraping and automating any part of its services." Its earlier scores are not shown. It returns when a person reads its pages in a browser.
A review of the other rows read from MoonPay's and Ramp's sites, after their off-ramp rows were held for the same terms.
Yes, see CC-2026-0142: Ramp Network (#4, 70) leaves on-ramps and its list in each of the 27 countries, with MoonPay.
The 139 earlier corrections, CC-2026-0001 to CC-2026-0140, are printed in full, with the same anchors, on the corrections archive. CC-2026-0080 was never issued: a build check uses that number as its test case.