We may earn a commission from operator links. This never affects our ratings.
BTC$75,240ETH$2,315USDT$1.00SOL$85LTC$84DOGE$0.16BTC$75,240ETH$2,315USDT$1.00SOL$85LTC$84DOGE$0.16
Home/Methodology

Methodology

We rank on seven categories, weighted toward the ones that actually matter when you try to leave with your money. The overall score is a weighted average of seven sub-scores, each on a ten-point scale, calculated automatically at build time. Every sub-score is visible on each review page, so you can see which category dragged the headline rating up or down.

The formula

CategoryWeightWhat it covers
Crypto payments20%Coin breadth, network options, deposit minimums, withdrawal speed, KYC thresholds.
Welcome bonus20%Headline match, rollover, game weighting, max bet limits, expiry, expected value to the player.
Odds quality15%Price versus industry on NFL and NBA main lines; vig across soccer main markets; prop depth.
Game variety15%Provider list, count of live titles, originals suite, live dealer depth, progressive presence.
Platform UX10%Desktop and mobile web quality, search, navigation, cashier clarity, load times.
Live betting10%In-play latency, market depth, cash-out availability, live streaming coverage.
Customer support10%Chat hours, median first-response time, agent competence on bonus and payment issues.

Why those weights

Payments is tied for the top weight because at a crypto casino, the thing most likely to go wrong and also the thing most worth complaining about is a withdrawal that does not arrive. Welcome bonus shares the top weight because the headline offer is also the single element most capable of destroying player EV when its rollover is set wrong. Odds quality is weighted lower than payments, which will feel backwards to sportsbook specialists; the rationale is that the sample of reliable operators with competitive pricing is small enough that the difference between 8 and 10 on odds matters less than the difference between 6 and 10 on payments.

Testing process

  1. Register a fresh account with a test email and a real IP (no VPN), from a supported jurisdiction.
  2. Deposit in three coins across at least two networks: USDT TRC-20 is always one; BTC or ETH is always another. Log the credit time for each.
  3. Place at least 20 real-money wagers across sports, slots, and live dealer (where offered).
  4. Request a withdrawal before the balance is fully run down. Time from click to on-chain confirmation.
  5. Contact support twice: once with a simple question, once with a bonus-related question. Log median first-response and resolution quality.
  6. Read the full terms and conditions for every active promotion. Note any max-bet rules, game exclusions, or geo limits not prominent on the landing page.
  7. Cross-check the licence number on the issuing authority's register. If the register does not confirm the licence, the score takes a half-point penalty until the operator reconciles it.

What the scoring does not capture

It does not capture jurisdiction-specific friction. A Stake account in Brazil behaves differently from a Stake account in Canada; we test from one jurisdiction (central Europe) and flag known exceptions in the review text rather than averaging them into the score. It also does not capture stake-size effects. A $100 player and a $10,000 player get different KYC handling, different risk reviews, and different bonus terms at every operator we tested; the scores reflect mid-stakes play, roughly $50 to $1,000 per session.

Re-test cadence

Every brand is re-tested quarterly. Scores change when the product changes, not on a schedule. If a site launches a new welcome bonus, the welcome bonus sub-score is re-scored within a week of launch. Payment and support sub-scores are updated whenever a reader flags a material regression and we confirm it in a spot check. The master review dates are stamped on each page.

Where this methodology can be wrong

Two known limitations. First, withdrawal speeds can regress across a quarter in ways that do not show up until a large sample accumulates; our ten-cashout sample is enough to catch a median shift, not every tail event. Second, bonus EV math assumes slots RTP is what the operator publishes; we do not run independent RTP tests, which would require far larger samples than our testing supports. Readers who spot a discrepancy between our score and their own experience are invited to email editorial.

Image sourcing

Operator screenshots shown on review pages were sourced from operator websites and third-party review sites (Wizard of Odds, CCN) where direct capture was blocked by Cloudflare. Each screenshot is captioned with the source site and the date captured. Where the operator logo appears, it remains the property of the operator.