Stake Dice: 99% RTP, Adjustable Edges and Provably Fair Math

You see a slider from 0 to 100 with two regions: 'roll under' and 'roll over'. Set your target — for example, roll under 50.50 — and the game generates a random number from 0 to 99.99 (uniform distribution). If the rolled number is under your target, you win; if over, you lose. The payout multiplier is calculated automatically based on your win chance and the 1% house edge: payout = 0.99 / (win chance / 100).

Set a 50.50% win chance and the payout is 1.98x — slightly less than the 2x that would be 'fair' (the 1% gap is the house edge). Set a 9.90% win chance and the payout is 10x. Set a 0.99% win chance and the payout is 100x. The math is uniformly transparent.

Stake Dice's RTP is 99% across all win chance settings — the 1% house edge is baked into the payout formula and doesn't change with your bet sizing or strategy. This is one of the highest RTPs on the platform, tying with Crash, Mines, and Plinko in the originals tier.

Over an infinite number of rolls, you'd lose exactly 1% of your turnover. Over a session of 1,000 $1 rolls, expected loss is $10. Variance around that depends entirely on what win chance you've selected — a 50% win chance produces tight clustering around -$10, a 1% win chance produces enormous variance with rare big wins.

High win chance (50%+) means you win most rolls but multiply your bet by less than 2x. Long sessions feel smooth — wins, wins, occasional losses, slow drift downward at 1%. Low win chance (1–5%) means you lose most rolls but big wins multiply by 20–100x. Long sessions feel brutal until a big hit.

Identical EV either way. The choice is purely variance preference. Players who want consistent results pick high win chance; players hunting for big multipliers pick low. There's no 'smart' choice — the math is constant.

Dice is the simplest game to verify because the output is a single number derived from server seed + client seed + nonce via a public hash function. Every roll publishes its server seed and nonce; you can recompute the result yourself in any programming language.

If you suspect a streak is rigged, you can verify every roll in seconds. The math has been independently audited for years and the algorithm is open. Run a verification script over 100,000 rolls and you'll see a textbook uniform distribution. See our provably fair guide for the verification walkthrough.

Dice is the game most associated with martingale because 50% win chance roughly doubles your bet. The pitch: lose a $1 bet at 50%, double to $2, lose, double to $4, etc. — eventually you win and recover all losses plus $1 profit.

It fails for the same reason it always fails: long losing streaks happen, bet sizing grows exponentially, and a streak of 10+ losses requires a $1,024 bet that may exceed table limits or your remaining bankroll. The 1% house edge ensures you bleed out long-run regardless of progression. Fixed bet sizing is the only mathematically defensible approach.

Stake's Dice auto-bet is the most configurable on the platform — set win chance, bet size, win/loss adjustments, stop conditions, and run thousands of rolls per session. For grinding wagering volume, Dice auto-bet at 49.5% win chance (low variance, 99% RTP) is one of the most efficient tools available.

Configure stop-loss and stop-profit aggressively — even with high RTP, variance can produce 20%+ drawdowns within a session. Set walk-away conditions and let the math play out. See our bankroll management framework for the broader context.

Because Dice has a 99% RTP and contributes 100% toward all promotional volume, it's mathematically the most efficient game for accumulating rakeback and cashback. Every dollar wagered generates 1% effective loss + whatever rakeback rate your tier earns + whatever cashback your tier earns on net negative weeks.

For high-tier VIPs, the combined rakeback + cashback can push the effective house edge to near-zero or even slightly positive on Dice. This is one of the few games where serious volume players can play mathematically close to break-even. See our rakeback vs cashback comparison for the math.