Variance in Casino Games

For any bet, the mathematician lists every possible result, notes how far each one sits from the average return, squares that distance and weights it by the result’s probability. Adding those weighted squares together gives the variance of a single round, expressed per unit staked.

Because the distances are squared, rare large wins dominate the figure. A jackpot paying thousands of times the stake pushes variance up far more than a steady stream of small wins ever could.

Players rarely see variance itself, since its units are squared euro and mean little to anyone. What studios and analysts usually quote instead is the standard deviation, which is simply the square root of the variance and comes back in the same units as the stake. Over many rounds the variance grows in proportion to the number of bets, so the standard deviation grows with the square root: a hundred spins carry ten times the spread of one spin, not a hundred times. That is why the house edge, which grows in a straight line, eventually outweighs the swings, though only across a volume of play far beyond any single evening.

The result is a property of the game’s maths, not of luck.

European roulette is the clearest worked example, because every bet on the single-zero wheel carries the same house edge of 2.70% yet a very different spread. The figures below were calculated by the desk from the 37 pockets and the standard payouts, per €1 staked on one spin.

On slots, variance is built into the paytable and the reel strips, and it is fixed by the studio when the maths model is designed. Most providers do not publish a numerical variance or standard deviation for their games.

What the player sees instead is a volatility label, often low, medium or high, or a scale of lightning bolts or stars on the game’s information screen. There is no industry standard behind those labels, so one studio’s “high” need not match another’s, and the desk treats them as a rough guide rather than a measurement.

Variance also matters with bonuses. A wagering requirement asks the player to stake a set amount before withdrawal, and a high-variance game makes it more likely that a balance runs out before that target is reached, even though the expected loss on the turnover is the same as on a calmer game with an equal RTP.

Hit frequency, the share of rounds that return anything at all, is a related figure some studios do publish. It hints at variance but is not the same thing, because a game can hit often with tiny wins and still hide a very large top prize.

The most common mix-up is with RTP. RTP is the average percentage of stakes a game returns over an enormous number of rounds, while variance measures how far real sessions stray from that average along the way. A 96% RTP on €500 of play implies an average return of €480 across many such sessions, yet in a high-variance game a single session can finish far above or far below that figure.

Volatility is the second confusion, though here the difference is mostly vocabulary. In casino writing the two words are used almost interchangeably, with volatility as the informal label and variance as the formal statistic underneath it.

The third is the belief that variance “evens out” in the short term, so that a cold machine is due to pay. Each spin from a certified random number generator is independent of the last, and past losses do not raise the chance of the next win.

Variance only describes the spread; it never predicts the next result.