DICE PROVABLY FAIR VERIFIER
One float, one roll: crypto Dice maps four HMAC bytes onto the 0.00–100.00 scale, which makes it the easiest game of all to audit. Enter a revealed seed pair and a nonce and watch this page land on the identical roll.
Dice ships pre-selected above. Rolls come from the standard construction HMAC-SHA256(server seed, "clientSeed:nonce:cursor"), checked against reference vectors for Stake and Shuffle; for anywhere else, confirm the scheme in the operator table first.
Byte-To-Roll Steps
If you only ever audit one casino game, make it Dice. There is no board, no path, no animation — just a slider, a target, and a two-decimal number between 0.00 and 100.00. That minimalism means the whole fairness argument fits in two lines of arithmetic, and a mismatch has nowhere to hide.
HOW DICE RESULTS ARE CALCULATED
Dice consumes a single random float from the HMAC stream and scales it to the 0.00–100.00 range:
- Compute
HMAC-SHA256(serverSeed, "clientSeed:nonce:0")and take the first 4 bytes as a float in [0,1). - Apply
roll = floor(float × 10001) / 100→ a value from 0.00 to 100.00.
Reference check: seed pair some server seed / some client seed at nonce 1 produces the float 0.59192619, i.e. a roll of 59.19. If the tool ever printed anything else, the port would be broken — it is pinned to the published test vectors.
TARGETS, WIN CHANCE AND PAYOUT
The roll itself is only half of Dice; the other half is the target you set before it. Betting under 25.00 wins on roughly 25% of rolls; over 75.00 likewise. The payout adjusts continuously as multiplier = 99 / win chance (%), so under-25 pays about 3.96x and a coin-flip-ish under-49.50 pays 2.00x. That 99 in the numerator (rather than 100) is the entire house edge — 1%, applied identically at every target. Verification only concerns the roll: the target, chance and payout are all visible on your bet slip, so once the roll reproduces, the settlement is a matter of reading the slider.
WHY VERIFY DICE?
Dice invites streak-hunting: after six losses in a row at 50% it is natural to suspect the roll generator is leaning on you. Auditing a losing run is the antidote — recompute each roll from the seed pair the casino committed to before the session and confirm every number was already locked in. What the check proves is narrow but decisive: the sequence you experienced was fixed before your first click, so no per-bet manipulation occurred. What it cannot do is make 99/chance into 100/chance — the edge calculator shows what that 1% costs over volume, and the provably-fair guide explains why the commitment itself is trustworthy.
Related Resources
Play Dice: Stake Dice
Strategy: Dice Strategy Guide — including what the best dice strategy actually is (and why none beat the edge)
Casinos where every roll is checkable: ranked provably-fair operators