Seven bonus clauses crypto casinos use to seize winnings, why sticky bonuses lock your cash, and how automated term auditing catches them before you deposit.
August 18, 2026
The data behind the score
Everything We Measure, And What We Refuse To Claim
What we read, and what it produces
Every operator is measured against the same sources, and none of them are sources an operator controls. That is the whole reason the number is worth anything.
What we are watching right now
Bets read from each operator's own public feed, counted by us. Return is what players actually got back across every game we saw, not a figure the operator supplied.
| Casino | Games | Studios | Bets seen | Observed return | Operator states | Difference | Biggest win |
|---|---|---|---|---|---|---|---|
| Roobet | 5,656 | 27 | 1,912,978 | 97.9% ±2.8% | not published | not published | 5,000x |
| Winna | 1,611 | 33 | 963,621 | collecting | not published | not published | 9,900x |
| Shuffle | 3,306 | 0 | 608,598 | collecting | not published | not published | 5,951x |
| Stake.com | 3,907 | 408 | 174,631 | collecting | 97.0% | not published | 80,085x |
Observed return is uncorrected. These feeds are the operators' own marketing surfaces and the selection leans toward winning rounds, so the figure is biased upward by an amount we cannot measure and that differs between operators. Where it exceeds 100% we label the feed curated, because no game returns more than it takes. Rolling 7 days. A return figure appears once an operator passes 1,000 observed bets, because below that the average is decided by whichever rare large win happened to land in the window. The range after each figure is the 95% interval. These feeds are rolling windows the operators publish themselves, so this is a sample of their play and not all of it.
Observed return, game by game
What players actually got back on the games we saw most, next to what the operator says the game returns. The second column is only filled where an operator publishes a house edge at all.
| Game | Where | Bets seen | Observed | Operator states | Difference | Biggest |
|---|---|---|---|---|---|---|
| Dice | Roobethousegames | 482,324 | collecting | not published | 4,950x | |
| Kenocurated feed | Winnainternal house games | 291,292 | collecting | not published | 800x | |
| Limbocurated feed | Winnainternal house games | 162,403 | collecting | not published | 9,900x | |
| Plinkocurated feed | Winnainternal house games | 140,386 | collecting | not published | 2,000x | |
| Keno | Shuffleunknown | 92,901 | 97.6% ±3.5% | not published | 450x | |
| Dicecurated feed | Winnainternal house games | 81,535 | collecting | not published | 9,900x | |
| Dice | Shuffleunknown | 64,657 | 98.1% ±1.1% | not published | 100x | |
| Fist of Destruction Megamultiplier | Roobethacksaw | 52,417 | collecting | not published | 918x | |
| Blackjack | Shuffleunknown | 45,001 | 97.9% ±0.9% | not published | 26x | |
| Limbo | Shuffleunknown | 43,750 | 98.2% ±7.0% | not published | 1,200x | |
| Wanted Dead or a Wild | Roobethacksaw | 39,665 | collecting | not published | 4,200x | |
| Wheel | Roobethousegames | 38,369 | 97.5% ±3.6% | not published | 50x |
Rolling 7 days, ordered by how much play we captured. Observed return is uncorrected for feed selection and biased upward by an unknown amount, so treat a single game's figure as a reading of the feed rather than a measurement of the game. The same play grouped by studio reaches a statable figure much sooner, because a proportion converges faster than a heavy-tailed mean. Every game as JSON.
What we can attribute on chain
Money moving inside the games is public at every casino we read. Money moving out to players is private at all of them, so payout figures have to come from the chain, and that means knowing which addresses belong to whom.
| Casino | Chain | Addresses | Given to us | Hot wallets | Deepest inference |
|---|---|---|---|---|---|
| Wild.io | ETH | 3 | 3 | 0 | 0 hops |
| Winna | BSC | 1 | 1 | 1 | 0 hops |
| Shuffle | ETH | 1 | 1 | 2 | 0 hops |
| Wild.io | BSC | 1 | 1 | 0 | 0 hops |
| Rainbet Casino | TRON | 1 | 1 | 1 | 0 hops |
| Wild.io | SOL | 1 | 1 | 0 | 0 hops |
| BetFury | BSC | 1 | 1 | 1 | 0 hops |
| BC.Game | TRON | 1 | 1 | 1 | 0 hops |
| Stake Casino | ETH | 1 | 1 | 5 | 0 hops |
| Rollbit | BTC | 1 | 1 | 0 | 0 hops |
| Cloudbet | LTC | 1 | 1 | 0 | 0 hops |
| Wild.io | TRON | 1 | 1 | 0 | 0 hops |
| BetFury | SOL | 1 | 1 | 1 | 0 hops |
| Duelbits | TRON | 1 | 1 | 1 | 0 hops |
| BC.Game | BTC | 1 | 1 | 1 | 0 hops |
| Shuffle | SOL | 1 | 1 | 2 | 0 hops |
| Duelbits | LTC | 1 | 1 | 0 | 0 hops |
| Rainbet Casino | BTC | 1 | 1 | 1 | 0 hops |
| Gamdom | TRON | 1 | 1 | 1 | 0 hops |
| Rollbit | ETH | 1 | 1 | 3 | 0 hops |
| Shuffle | BTC | 1 | 1 | 0 | 0 hops |
| Gamdom | BSC | 1 | 1 | 2 | 0 hops |
| Stake Casino | LTC | 1 | 1 | 0 | 0 hops |
| BC.Game | BSC | 1 | 1 | 2 | 0 hops |
| Cloudbet | ETH | 1 | 1 | 2 | 0 hops |
| Gamdom | ETH | 1 | 1 | 3 | 0 hops |
| BC.Game | SOL | 1 | 1 | 1 | 0 hops |
| Shuffle | LTC | 1 | 1 | 0 | 0 hops |
| Gamdom | BTC | 1 | 1 | 0 | 0 hops |
| Rainbet Casino | BSC | 1 | 1 | 1 | 0 hops |
| Duelbits | BTC | 1 | 1 | 0 | 0 hops |
| BC.Game | LTC | 1 | 1 | 0 | 0 hops |
| Wild.io | LTC | 1 | 1 | 0 | 0 hops |
| Duelbits | BSC | 1 | 1 | 1 | 0 hops |
| Rollbit | SOL | 1 | 1 | 1 | 0 hops |
| Gamdom | LTC | 1 | 1 | 0 | 0 hops |
| Rainbet Casino | SOL | 1 | 1 | 1 | 0 hops |
| Wild.io | BTC | 1 | 1 | 0 | 0 hops |
| BC.Game | ETH | 1 | 1 | 8 | 0 hops |
| Winna | TRON | 1 | 1 | 1 | 0 hops |
| BetFury | LTC | 1 | 1 | 0 | 0 hops |
| Stake Casino | TRON | 1 | 1 | 1 | 0 hops |
| Winna | ETH | 1 | 1 | 1 | 0 hops |
| Cloudbet | BSC | 1 | 1 | 2 | 0 hops |
| Cloudbet | TRON | 1 | 1 | 2 | 0 hops |
| BetFury | ETH | 1 | 1 | 1 | 0 hops |
| Stake Casino | BSC | 1 | 1 | 2 | 0 hops |
| Cloudbet | BTC | 1 | 1 | 0 | 0 hops |
| Winna | BTC | 1 | 1 | 0 | 0 hops |
| Winna | LTC | 1 | 1 | 0 | 0 hops |
52 attributed addresses. Every address has to trace back to one the operator handed us. Totals built on them are a floor, never an operator's real position, because wallets we have not found are invisible. See every address and how we know.
Latest from the blog
What we find while auditing operators, and what changes in the data behind the scores.
Most casino review sites are paid by the operators they rate. Automated term auditing, 2026 US tax calculations, and payout speed benchmarks instead.
August 12, 2026
I tested five offshore casinos with $50 each and timed every cashout, with KYC triggers, state availability and fees recorded for each.
August 7, 2026
How we score.
Scores are computed by a published algorithm. No commercial arrangement is an input to one.
