Count winning results, not the target
A Dice roll has 10,000 possible results from 0.00 to 99.99. Roll Under 49.50 wins on 0.00-49.49: 4,950 results, a 49.50% chance. Roll Over 50.49 wins on 50.50-99.99: also 4,950 results. That's why the Over target for the same chance isn't simply 100 minus the Under target: the tie value always loses, so it's left out on both sides.
From win chance to payout
Payout = (100% − house edge) ÷ win chance. The table uses a 1% edge. At a 2% edge, like BCH.games, every payout is about 1% lower: 50% pays 1.96× instead of 1.98×.
Chance of a losing streak
Each roll is independent, so the chance of losing k rolls in a row is (1 − win chance)ᵏ.
| Win chance | 3 losses in a row | 5 in a row | 10 in a row |
|---|---|---|---|
| 90% | 1 in 1,000 | 1 in 100,000 | ~1 in 10 billion |
| 49.5% | 1 in 7.8 | 1 in 30 | 1 in 927 |
| 10% | 1 in 1.4 | 1 in 1.7 | 1 in 2.9 |
A 1-in-927 streak sounds rare, but over 1,000 rolls at 49.5% there's about a 41% chance of seeing at least one run of 10 losses.
About these numbers. The illustrative model rolls one of 10,000 equally likely results from 0.00 to 99.99, and a result equal to your target loses. The BCH.games profile uses rules captured on October 25, 2026, including its nominal 2% house edge. Other operators may use a different range, edge or rounding. How the Dice models work.
