What the Dice comparison shows
Common crypto Dice bets win when a roll falls below or above a target. The chance determines how often a bet wins; the payout determines how much is returned. At an exact 50% win probability, a 2% house edge gives a 1.96x gross payout, while a 1% edge gives 1.98x. Gross payout includes the stake. Different roll grids and rounding can make a nominal percentage differ slightly from the exact probability.
The comparison shows each operator model’s payout and return at the chance setting you choose. It does not rank casinos. Account terms, regional availability, withdrawals and support are separate questions that a payout formula cannot answer. Operators are listed alphabetically, so position is not a recommendation.
Dice house edge by casino
| Operator | Model return | Model house edge | Gross payout | Evidence |
|---|---|---|---|---|
| BC.Game (Dice98) | 98.00% | 2.00% | 1.96x | Reference estimate; Verified roll replay |
| BCH.games | 98.00% | 2.00% | 1.96x | Checked payout rules; Verified roll replay |
| Duel.com (base tier) | 99.90% | 0.10% | 1.9982x | Reference base-tier estimate; Verified roll replay |
| Duelbits | 99.01% | 0.99% | 1.9802x | Announced multiplier formula; Reference |
| Stake | 98.99% | 1.01% | 1.98x | Public client formula; Reference |
Why Stake shows 98.99%, not 99%. Its supported mapping has results from 0.00 to 100.00. Under 50 wins on 5,000 of 10,001 values: about 49.995%. The client formula pays 1.98x for the nominal 50% setting, so calculated return is about 98.99%. The calculation counts the actual 32-bit mapping. The Duel.com base-tier estimate prices the 10,001-value grid and rounds the payout to four decimals, giving about 99.90% return. This remains a model estimate beyond the captured target.
Why BC.Game can show 97.99%. The Dice98 estimate rounds payouts down to four decimals, consistent with the captured examples. At a 90% setting, 0.98 ÷ 0.9 is about 1.08889x; rounding down gives 1.0888x and 97.992% return. This illustrates the model convention, rather than confirming every live target. Duelbits uses the announced formula 100 ÷ (chance percent × 1.01); its displayed multiplier is rounded to four decimals here.
What the difference costs
Expected loss = total wagered × effective house edge. Total wagered means every stake placed, not just the initial deposit. At a fixed return, changing win chance changes the swings rather than the cost per unit wagered; grid effects and payout rounding can change the effective edge slightly.
| Total wagered | 2% edge | 1% edge | 0.1% edge |
|---|---|---|---|
| $100 | $2.00 | $1.00 | $0.10 |
| $1,000 | $20.00 | $10.00 | $1.00 |
| $10,000 | $200.00 | $100.00 | $10.00 |
Expected loss is an average, not a prediction for one session. If all 1,000 independent bets of one unit can be funded, a roughly 50% chance setting gives about a 25% chance of finishing ahead at a 2% edge, about 36% at 1%, and about 49% at 0.1%. These binomial examples exclude early stopping and funding limits. Use the Dice simulator to explore the wider spread under a selected bankroll.
Compare at the same win chance, not the same target
Operators use different result ranges, including 0.00–99.99 and 0.00–100.00. Copying a target of 50 therefore need not copy the exact probability. Start with the same nominal win chance, then read the model’s grid and payout rules. The comparison does not claim the actual probabilities are identical.
Roll Over and Roll Under can describe equivalent winning counts when their thresholds are adjusted for the grid. A visual target such as Over 75 is not universally interchangeable with Under 25. The shared comparison is limited to 0.01–97.99% to stay within the supported BCH.games settings; this is not a claim that every other operator uses those same limits.
Low edge, cashback and rakeback
Cashback or rakeback can return part of wagering costs later under an operator’s terms. It does not change the probability of a roll. Rebates can depend on VIP level, eligibility or stake size. This comparison uses gross game return before cashback. Duel.com’s model covers the base tier; larger stakes can use different payout tiers.
Payout comparison is not fairness proof
A payout describes the return for a winning bet. Provably fair evidence tests whether a recorded result follows from disclosed seeds and the supported algorithm. BC.Game, BCH.games and Duel.com have Verified roll replay adapters within their documented scopes; Stake and Duelbits are Reference. Neither status confirms a complete payout profile or casino safety. Choose the matching Dice verifier for your recorded bet.
Related: Compare Plinko · Dice calculator · Dice strategy · House edge explained · Methodology
