Recompute any Stake-style result yourself β one HMAC-SHA256 engine for Stake, Stake.us, Rainbet, Roobet, BC.Game, Shuffle & more. Dice, Limbo, Plinko, Mines, Keno.
This tool computes everything locally via the Web Crypto API (HMAC-SHA256). There is no server call β your server seed, client seed and nonce never leave your device. View the math Β· it's the same code we test against Stake's own calculator.
Prefer a single-game page with the exact formula spelled out? Open the one you need:
Dice verifier Β· Limbo verifier Β· Plinko verifier Β· Mines verifier Β· Keno verifier
Before a bet, the casino shows you a hashed server seed (a commitment it can't change later). Your bet also uses your own client seed and an incrementing nonce. The result comes from:
HMAC-SHA256( key = server_seed , message = "client_seed:nonce:round" )
That 32-byte digest is sliced into numbers in [0,1) and mapped to the game outcome. Once the casino reveals the server seed, you can recompute every past round exactly β which is what this tool does. If the recomputed result matches what the game showed, that round was honest.
These are the real mappings, the same ones we verified bit-for-bit against Stake's own /provably-fair/calculation page.
| Game | Floats | Mapping |
|---|---|---|
| Dice | 1 | floor(f Β· 10001) / 100 β 0.00β100.00 |
| Limbo | 1 | 16777216 / (floor(f Β· 16777216) + 1) Β· 0.99 β crash multiplier (2Β²β΄-based β not the 1e8 many verifiers get wrong) |
| Plinko | rows (8β16) | each float β Left/Right, bucket = number of Rights |
| Mines | = mines | sequential draw without replacement from 25 tiles (draw order) |
| Keno | 10 | draw 10 without replacement from pool 1β40 |
β οΈ For Plinko and Mines, set the row/mine count to exactly what the round used β a different setting gives a different (but still correct) result.
Any casino that uses the Stake-style HMAC-SHA256 engine for per-bet games: Stake, Stake.us, Rainbet, Roobet, BC.Game, Shuffle and most of their clones share the exact same math. That's why one verifier covers them all. Crash in house-seed mode (a public seed chain instead of your client seed) works differently β see our dedicated Blaze Crash verifier and Aviator verifier.
Yes. The calculation runs entirely in your browser using the Web Crypto API. Nothing is sent to any server, logged, or stored. You can disconnect from the internet and it still works β and you can read the source of this page to confirm it.
No β and be very suspicious of anyone who claims to. While the server seed is still hashed (unrevealed), nothing can be predicted. This tool only checks past rounds after the seed is revealed. Any "predictor" that promises the next multiplier is a scam.
It means that single round was generated honestly from the committed seeds β the casino didn't tamper with that specific outcome. It does not tell you the house edge or whether the game is a good deal. Fair β cheap. Our audits cover the house edge separately.
Because they license or copy the same provably-fair implementation Stake pioneered. The HMAC-SHA256 byte stream and the game mappings are identical, so the same code recomputes results across all of them.
First check the obvious: the server seed must be the revealed/unhashed value, the nonce must match, and for Plinko/Mines the row/mine count must be exactly what the round used. A stray space in a pasted seed is the most common cause. If it still won't reconcile with correct inputs, that's exactly the kind of case worth reporting β see our methodology.
Formulas verified bit-for-bit against Stake's reference calculator (Aug 2026). Open methodology: /methodology/. Β© FairPlay Audit β independent, free, open-source.