Ratings Methodology
We score casinos from 1 to 10 in tenths of a point. This page sets out what an casino has to clear before we will score it at all, how the score is weighted, and what evidence sits behind each criterion.
One principle runs through all of it. Measured evidence outweighs stated evidence, and stated evidence outweighs reported evidence. An casino that publishes a withdrawal threshold has made a commitment we can hold it to. One that stays silent has reserved the right to decide after you have already deposited, and it scores accordingly.
What Our Confidence Labels Mean
Every figure we publish is tagged with how we know it, and the tag sits next to the number rather than in a footnote. You never have to guess whether something was measured or repeated from a forum.
| Label | Means | Evidence we hold |
|---|---|---|
| Measured | We produced it ourselves | Transaction hash and four timestamps |
| Casino states | Published by the casino | The clause quoted verbatim, its URL, and the date pulled |
| User reports | Aggregate reports, unconfirmed | None. Treated as hearsay and worded that way. |
A reported figure never counts toward a score. It can appear in the text where it is the only information available, but it is labelled and hedged, and it carries no weight in the rating.
Inclusion Criteria
An casino has to clear four gates before we score it. These are pass or fail rather than weighted, and an casino that fails one is recorded in the register with the reason rather than quietly dropped.
1. A verifiable licence (pass or fail)
Why it matters. A licence number that does not resolve to a registry entry tells you the casino is willing to misrepresent its regulatory position, which is the cheapest possible lie to check.
How it is measured. We look the number up in the issuing registry and record the date we checked. A footer claim with no matching entry fails.
2. A completed withdrawal (pass or fail)
Why it matters. Everything else is theoretical until money has actually come out. This is the single most common failure mode in this market and the least reported.
How it is measured. At least one withdrawal requested from a funded account and confirmed on chain, with the transaction hash recorded. Held or refused withdrawals fail the gate and the outcome is published.
3. Published terms (pass or fail)
Why it matters. Terms that cannot be read before depositing cannot be relied on afterwards, and an casino that hides them has removed your ability to make an informed decision.
How it is measured. Terms must be reachable without an account and must state the conditions under which verification can be demanded and funds withheld.
4. Responsible gambling provision (pass or fail)
Why it matters. Deposit limits, time-outs and self-exclusion are the controls that matter most to the people most at risk, and their absence is a reliable signal of how an casino views its users.
How it is measured. We check the tools exist, are reachable from the account area, and actually apply when set.
Rating Criteria
Casinos that clear all four gates are scored against six weighted criteria. The weighting is deliberately front-loaded onto things that require evidence to assess, rather than things visible from a homepage.
1. Settlement speed (25%)
Why it matters. How long it takes to get paid is the thing most players care about and the thing least often measured. Published payout times blend the casino's own approval queue with network confirmation, which hides the part the casino controls.
How it is measured. Timed withdrawals across four separate timestamps, scored on the median approval delay rather than total elapsed time. Every measurement carries its transaction hash.
2. Verification and privacy (20%)
Why it matters. A deferred identity check is only an advantage if you know when it arrives. The risk is not being asked for documents, it is being asked while your balance is held.
How it is measured. What the terms reserve, what a real withdrawal actually triggered, and whether the two agree. An casino whose practice is stricter than its terms scores below one where they match.
3. Terms clarity (15%)
Why it matters. A published threshold is a commitment. Silence is an option the casino holds and you do not, and it is exercised at the point you are trying to withdraw.
How it is measured. Whether verification thresholds, withdrawal limits and void conditions are stated as numbers. Discretionary language scores down even where the casino behaves well in practice.
4. Licence and recourse (15%)
Why it matters. Licences in this market differ enormously in what they require. Some run a complaints process a player can use, and others impose no gambling-specific obligation at all.
How it is measured. The regime, the registry entry, and whether a dispute mechanism exists that has resolved cases. A Costa Rica business registration scores near zero because it is not a gambling licence.
5. Chain and asset support (15%)
Why it matters. Network choice decides your fees and your settlement time, and an casino supporting only expensive chains costs you money on every transaction.
How it is measured. Networks and assets accepted, minimum deposit and withdrawal, and fees charged on top of network cost. Taken from casino documentation and confirmed against our own transactions.
6. Support responsiveness (10%)
Why it matters. Support matters most during a withheld withdrawal, which is exactly when casinos are slowest to respond.
How it is measured. Timed contacts on a real account question, scored on median first response and whether the answer resolved the issue.
How We Time a Withdrawal
We record four timestamps rather than one, because a single payout figure conceals who caused the delay. The gap between approval and broadcast is the casino’s own queue and it is the only part it actually controls.
| Timestamp | What it marks | Who controls it |
|---|---|---|
| Requested | Withdrawal submitted in the casino interface | You |
| Approved | Casino marked it approved | Casino |
| Broadcast | Transaction reached the mempool | Casino |
| Confirmed | Confirmed on chain | Network |
A measurement is published only after the confirmation time has been independently re-derived from the transaction hash. Testers are paid on verified measurements rather than submitted ones, which removes the incentive to report a time that did not happen.
Failed withdrawals are recorded the same way successful ones are. An account frozen mid-withdrawal, a payout refused, or a document request that blocked a cash-out are the most useful outcomes we can report, and they are never dropped.
Data Sources
- Casino terms. Snapshotted and diffed against the previous pull, so a changed clause is recorded with both values and the date. Quoted verbatim wherever a figure comes from them.
- Licence registries. Looked up directly rather than taken from a footer, with the check dated.
- Our own transactions. Deposits, withdrawals and support contacts from funded accounts, with on-chain proof attached.
- Block explorers. Used to re-derive confirmation times independently of anything the casino or the tester reports.
Community reports are read but never scored. They are useful for knowing where to look and they are not evidence, so a figure that exists only because a forum said so is labelled as reported and left out of the rating.
Provisional Scores
A casino we have not yet paid out from carries a provisional score rather than no score at all. It is computed from the criteria we do hold evidence for, renormalised across the weight actually covered, and capped at 7.0.
The cap is the point. Settlement speed is a quarter of the full weighting, so a score without it is missing the most important thing we measure. A provisional score tells you how a casino compares on what is knowable from its terms and its published figures, and nothing about whether it pays.
Provisional scores are marked. When a withdrawal is timed the cap lifts, settlement enters the calculation, and the score moves to reflect it.
What Each Score Means
| Score | What it means |
|---|---|
| 1 to 3 | Avoid. A failed gate, a withheld payout, or a licence claim that does not resolve. |
| 4 to 6 | Works, with a known problem. Usually slow settlement or terms that reserve wide discretion. |
| 7 to 8 | Solid. Pays reliably, but something is undisclosed or unmeasured. |
| 9 to 10 | Measured across multiple withdrawals, publishes its thresholds, and the terms match the behaviour. |
| marked provisional | No withdrawal timed yet. Capped at 7.0 and scored only on what its terms and published figures support. |
A score above 9 requires repeat measurements rather than one successful withdrawal. A single payout proves an casino can pay, not that it does so consistently.
On Statistical Normalisation
We do not normalise scores across casinos yet, because the register is too small for it to mean anything. Standardising a criterion against a sample of a handful of casinos produces confident-looking numbers with nothing behind them.
Once the register holds thirty casinos with measured settlement data, we will normalise the measured criteria against the sample and say so here on the date it changes. Until then the scores are weighted absolutes and should be read as such.
Corrections and Updates
Figures are never edited in place. A changed value is appended as a new dated record and the previous one stays in the record, so a score that moves can always be traced to the evidence that moved it.
Measured figures are re-tested rather than assumed to hold. An casino can change its payment processor without announcing it, which is why every published figure carries the date it was recorded and why an undated number is not evidence.
If you can show a figure here is wrong, we will correct it and date the correction. Our editorial standards cover how bylines, sourcing and commercial relationships are handled.
Commercial Relationships
Some outbound links to casinos are commercial and are marked in the page source. They do not affect a score, a measured figure, a ranking position, or whether an casino is covered at all.
Casinos with poor records stay in the register with poor records. A failed inclusion gate is published with the reason, which is the part of this that would be worth removing if commercial terms influenced it, and it is the part we will not remove.