Variance
The statistical measure of how much actual results deviate from the expected value — in crash games, high variance means long losing streaks and occasional large wins; low variance means frequent small wins and smaller swings.
Variance is the reason crash game sessions feel nothing like the mathematical expected value. You can play 20 rounds at 2x and bust 15 of them (despite a ~50% expected win rate) — this is not bad luck in a meaningful sense, it is normal variance.
High variance vs low variance targets
| Target | Win rate | Variance level | Streak risk |
|---|---|---|---|
1.20x | ~80% | Very low | Rare losing streaks |
2.00x | ~49% | Low-medium | Short streaks |
5.00x | ~19% | High | Long streaks common |
10.00x | ~9.7% | Very high | 10+ round streaks normal |
100x | ~1% | Extreme | 100+ round droughts expected |
All rows have the same expected value per bet. The variance level determines the experience of that EV — how smooth or volatile the ride is.
Why variance matters for bankroll
High variance requires a larger bankroll buffer. If you target 10x with a 9.7% win rate, you should expect a streak of 20+ consecutive busts to occur eventually. With a small bankroll, that streak busts you out before variance can revert toward EV.
The standard formula: minimum bankroll = 1 / win_rate × 10 base bets.
- At
50%win rate (2x target): 20 units minimum - At
10%win rate (10x target): 100 units minimum - At
2%win rate (50x target): 500 units minimum
Variance is not bias
A run of 10 consecutive busts does not make a win “overdue.” Each round is independent. The crash point has no memory of previous rounds. This is the gambler’s fallacy — and it is the primary cognitive error that leads players to override their strategy after a losing streak.
Related terms
- Volatility — used interchangeably in casual use, technically different
- Bankroll — must be sized to survive your chosen variance level
- Hit Rate — the win rate that determines variance level