provably fair · verifier · house-edge database
Recompute the bet. Don't take their word for it.
Paste the seeds a casino gave you and VerifyRoll rebuilds the result with the same maths they used. If the numbers match, the outcome was fixed before you bet. If they don't, you have proof.
- 01Commitment. The revealed server seed hashes (SHA-256) to the value the casino published before you bet.
- 02Derivation. HMAC-SHA256(server seed, client seed:nonce:cursor) is recomputed and mapped through the game’s own formula.
- 03Payout. For Plinko the recomputed bucket is matched against the published multiplier table for that row count and risk.
verifierstake scheme · 7 games
Loading verifier…
house edge · published RTPchecked 2026-09-23
| Game | Stake | Shuffle | BC.Game | Craze |
|---|---|---|---|---|
| Dice | 99% | 99% | 99% | — |
| Limbo | 99% | 99% | 99% | — |
| Crash | 99% | 99% | 99% | — |
| Plinko | 99% | 99% | 99% | — |
| Mines | 99% | 99% | 99% | — |
| Keno | 99% | 99% | 99% | — |
what fair does not mean
Provably fair proves the result. It does not prove the odds.
A perfectly verifiable Dice game can still pay 99% or 96% of what it takes. The scheme fixes the random number; the payout table decides what that number is worth. That is why VerifyRoll pairs the verifier with a house-edge database built from what operators publish, dated, and marked “—” where they publish nothing.
VerifyRoll is sponsored by Craze, which competes with the other operators in that table, so we show only their published numbers, dated, and our methodology explains how we check them.
tools · exact probabilityall tools
- Losing-streak oddsDice 49.5%: a run of 8+ losses appears somewhere in 1,000 bets with probability 87.99%.
- Mines survival3 mines, 5 safe picks: 49.57%, 1 in 2.02.
- Crash target oddsA round reaches 2× 49.50% of the time; the return is 0.99 per unit at every target.
- Plinko buckets16 rows, medium: exact RTP 98.9883%; the 110× edges are 1 in 32,768.
- Dice calculator49.5% pays 2.0000× under the 99% rule; the expected loss is 1% of the stake at every chance.
changelog
- 2026-09-27Verify by casino, per-game verifier pages and four calculator landing pages. A page for every operator in the directory (/verify/stake, /verify/shuffle and six more) stating whether its bets recompute here, where the seeds are and a worked example; for unsupported operators an honest "not yet" with a browser-side SHA-256 commitment checker. Fourteen Stake and Shuffle per-game verifier pages with the derivation, published RTP and a build-time example. A new Dice calculator (chance, multiplier, payout, exact share of the 10,001 rolls) and query-focused Mines, Crash and Plinko calculator pages; the existing tool URLs are unchanged.
- 2026-09-23Probability tools, game maths, casino directory and a seed vault. Four exact calculators (losing streaks by recurrence, Mines survival, Crash targets, Plinko buckets with a chart); a maths page per Original with the derivation and a build-time worked example; a directory stating which operators the verifier supports; and a browser-only vault for saved verifications with JSON export.
- 2026-09-23New identity, structured data and a stricter security policy. VerifyRoll gets its own typeface and layout instead of the shared template, share images generated per page, schema.org markup on every page, a nonce-based Content Security Policy and published about, methodology and contact pages.
questions
- What does VerifyRoll actually check?
- That the revealed server seed matches the hash published before play, and that recomputing HMAC-SHA256 over the seeds and nonce produces the result the casino showed you. Both run in your browser; nothing is uploaded.
- Which casinos does it work with?
- Any operator using the Stake-style scheme: HMAC-SHA256 of client seed, nonce and cursor over a committed server seed. Stake and Shuffle are confirmed. BC.Game uses a different construction and is not supported yet.
- Does a passing verification mean the game is fair?
- It means the result was fixed before you bet and was not changed afterwards. It says nothing about the odds. For that, read the RTP and house-edge table, which is the other half of this site.