Methodology
How we score crypto casinos
Every overall casino score uses the five criteria and fixed weights below. You can reproduce the arithmetic, inspect the editorial judgements behind it, and see the evidence standards for source checks and firsthand tests.
The model
Five criteria and what they weigh
The weights are fixed and identical for every operator. Trust carries the most because a fast payout is worthless if the operator disputes the withdrawal; support carries the least because good support cannot rescue a bad product.
| Criterion | Weight | What it covers |
|---|---|---|
| Trust & safeguards | 30% | Licensing, operating history, ownership transparency, dispute record, account security options, and how the terms treat a player who wins. |
| Payout experience | 25% | Observed time from withdrawal request to funds arriving, plus verification friction, stated limits, and whether the process behaves the same on a large win as a small one. |
| Game library | 20% | Breadth and quality of slots, live dealer, and originals; provider mix; and whether provably fair games are genuinely verifiable. |
| Bonus value | 15% | What an offer is actually worth once wagering, game weighting, max-bet caps, and cash-out limits are applied — not the headline number. |
| Support | 10% | Response time, channel availability, and whether an agent resolves a payment question or recites a script. |
Worked example: Stake
Nothing is withheld here. These are the five numbers, the five weights, and the result — which is the score shown everywhere else on the site.
| Criterion | Score | Weight | Contribution |
|---|---|---|---|
| Trust & safeguards | 4.7 | 0.30 | 1.410 |
| Payout experience | 4.9 | 0.25 | 1.225 |
| Game library | 4.9 | 0.20 | 0.980 |
| Bonus value | 4.3 | 0.15 | 0.645 |
| Support | 4.5 | 0.10 | 0.450 |
| Overall score | sum of contributions | 4.7 | |
Scores are held at full precision internally and rounded only for display, so two operators showing the same rounded figure can still be ordered correctly. Ties are broken by the trust sub-score.
Evidence standards
What a documented check requires
These stages describe the records a check needs, rather than completed tests for every operator. Each review should distinguish public-source research from retained firsthand observations and identify unfinished checks. A repeatable scoring formula cannot replace evidence for its inputs. Dated public-source datasets: licence register check and payment network sources.
- 01
Deposit
For a firsthand deposit check, retain the date, asset, network, amount, fees, and observed timing. Published cashier documentation is useful source evidence, but it does not prove that an account was funded or that those conditions applied to a real deposit.
- 02
Play
A hands-on product check needs a record of what was used and when. A fairness check must match the pre-bet commitment and the operator's specific calculation method. Checking a sample outcome does not prove that every round is fair or that a game has positive expected value.
- 03
Withdraw
An observed withdrawal needs a dated request and receipt record, amount, asset, network, fees, and verification state. Distinguish operator approval from blockchain settlement. Without a retained record, timing stays unmeasured; one test cannot establish a typical time or guarantee future payouts.
- 04
Challenge
Source reviews should identify the terms and regulator records actually checked, with dates and unresolved gaps. A support assessment should retain a dated interaction. Published policies and an observed response answer different questions, so neither should be presented as the other.
Independence
What money can and cannot buy here
We earn a commission when a reader opens an account through some of our links. That is the business model, it is disclosed on every page that carries such a link, and pretending otherwise would be the first thing worth distrusting about us.
Every published operator has an active affiliate link. That commercial threshold decides what can be published, while the ranking itself is calculated from the same fixed model for every operator. No agreement can purchase a position, score, badge, or sentence.
Commercial links are labelled with sponsored link attributes and disclosed in visible text. Where we have not verified a fact, we leave it blank rather than filling it with something plausible.
- No paid placements
Position is a consequence of the score. There is no mechanism to buy one.
- Same criteria for everyone
One model, five fixed weights, applied to all 7 published operators.
- Commercial links disclosed
Affiliate links are visibly disclosed and marked sponsored and nofollow.
- Gaps stated, not filled
Unverified facts are shown as unconfirmed rather than guessed at.
Questions
The awkward ones
You earn commission from some of these operators. How is that compatible with independence?
Every published operator has an active affiliate link, so readers should treat commercial selection and editorial scoring as separate questions. Commercial eligibility determines which reviews appear; it cannot buy a position, score, badge, or wording change. Every score decomposes into five sub-scores with fixed published weights, making the arithmetic checkable rather than merely promised.
Are the sub-scores objective measurements?
No. They are editorial judgements informed by the evidence shown on each review, including firsthand testing where recorded. The weighted formula makes the arithmetic reproducible, but does not independently validate its inputs. A payout observation is not a large-sample average; a review without a retained timing record should say it is unmeasured. Trust reflects documented licensing, history, and terms, not a guarantee that an operator is safe.
What is the difference between a source check and a firsthand test?
A source check records what an operator's documentation or a regulator's record says, with a link and check date. A firsthand test records a specific action we performed, with its date, conditions, result, and supporting evidence. Reading published withdrawal terms is not a withdrawal test. Neither kind of check establishes that every feature of an operator has been tested.
What would make you drop an operator entirely?
Confiscating balances on pretextual grounds, refusing withdrawals from winning players, systematically misrepresenting bonus terms, or operating without any credible licence. Delisting is a stronger signal than a low score and we reserve it for conduct rather than mediocrity.
How often is this re-checked?
Reviews are revised when material evidence or wording changes. Editorial revision dates, source-check dates, and firsthand-test dates describe different events and should remain distinct. Updating a page does not mean its payment tests or licence checks were repeated, and an older check does not establish that current conditions are unchanged.
Can an operator get its score changed?
By providing evidence that a fact we relied on is wrong, yes — and we will correct it and say so. By any commercial arrangement, no. Those are the only two routes, and they are not negotiable against each other.
Think a score is wrong? That is what publishing the components is for — tell us which one and why.