The number you are chasing is a measure of swing, not of profit
A player sits down at a roulette table with a fixed bankroll and wants to know how far a session can realistically move from the average before the wheel has even spun. That is what standard deviation answers. It does not tell you whether you will win. It tells you how wide the range of ordinary outcomes is around the expected result, and therefore how much of a cushion a given bankroll provides.
Roulette is a useful place to learn the calculation because every bet has a known probability, a known payout, and a known expected loss per spin. Once those three numbers are in place, the standard deviation falls out of a formula that works the same way for any bet on the layout.
Start with the expected value of one spin
Expected value is the average result per spin if the same bet were repeated an enormous number of times. For a single-number bet on a European wheel, the probability of winning is 1 in 37, because the wheel has 37 pockets and only one of them pays. A winning single-number bet pays 35 to 1. The expected value per unit wagered is therefore:
EV = (35 × 1/37) + (−1 × 36/37) = (35/37) − (36/37) = −1/37
That is a loss of about 0.027 units per unit wagered, which is the house edge on that bet under that wheel layout. The same arithmetic on an American wheel uses 38 pockets and a 1 in 38 winning probability, producing an expected loss of 2/38 per unit. The extra pocket is the entire reason the expected loss is larger on the American layout.
Expected value is not what a player experiences in an evening. A single-number bet loses one unit on 36 of 37 spins and wins 35 units on the other. The average is a small loss, but no individual spin ever produces that average. Standard deviation measures the distance between the individual outcomes and that average.
The variance formula for a single spin
Variance is the average of the squared distances from the expected value. For a bet with two possible outcomes it is calculated as:
Variance = (win outcome − EV)² × P(win) + (loss outcome − EV)² × P(loss)
For the single-number bet on a European wheel, the win outcome is +35 units and the loss outcome is −1 unit. With EV equal to −1/37, the calculation runs:
Variance = (35 − (−1/37))² × (1/37) + (−1 − (−1/37))² × (36/37)
The first term is (35 + 1/37)², which is approximately 1,296.7, multiplied by 1/37. The second term is (−36/37)², which is 1,296/1,369, multiplied by 36/37. The result is approximately 34.08 units squared. The standard deviation is the square root of that variance, which is about 5.84 units per spin.
That figure is the scale of a single spin's swing around its expected value. A one-unit single-number bet typically lands several units away from its average, because the payout is large and rare. The same calculation for an even-money bet produces a much smaller standard deviation per spin, because the outcomes are closer together.
Why the bet type changes the standard deviation
Different roulette bets carry different payout structures, and that changes both the house edge and the per-spin standard deviation. The comparison below uses a European wheel and one unit wagered on each bet.
| Bet | Winning probability | Payout | Expected value per unit | Standard deviation per spin |
|---|---|---|---|---|
| Single number | 1/37 | 35 to 1 | −1/37 | about 5.84 units |
| Split (two numbers) | 2/37 | 17 to 1 | −1/37 | about 4.02 units |
| Street (three numbers) | 3/37 | 11 to 1 | −1/37 | about 3.24 units |
| Even-money (red/black, odd/even, high/low) | 18/37 | 1 to 1 | −1/37 | about 0.999 units |
The expected value is identical across all four bets on a European wheel: one unit wagered loses 1/37 of a unit on average. What changes is the standard deviation. A single-number bet swings roughly six times harder per spin than an even-money bet, even though both lose at the same long-run rate. That is the practical meaning of volatility: it is the shape of the ride, not the direction of the destination.
Scaling from one spin to a session
Standard deviation scales with the square root of the number of independent spins when the same bet is repeated. If the per-spin standard deviation is s, then after n independent spins the standard deviation of the total result is s × √n.
As an example, suppose a player makes 100 single-number bets of one unit each on a European wheel. The expected total result is 100 × (−1/37), which is about −2.7 units. The standard deviation of the total is 5.84 × √100, which is 58.4 units. The session's ordinary range of outcomes is therefore far wider than its expected loss. A result of −60 units or +55 units is well within one standard deviation of the expected value, even though the expected value itself is a small loss.
The same 100 spins on even-money bets produce a standard deviation of about 0.999 × √100, or 10 units, around the same expected loss of about −2.7 units. The two sessions have the same long-run expectation but completely different short-run shapes. A player who budgets for the average and ignores the standard deviation has not budgeted for the session they will experience.
What the formula does and does not tell you
Standard deviation is a description of dispersion, not a prediction of the next spin. It allows a player to estimate how often a session result will fall inside a given band around the expected value, under the assumption that each spin is independent and identically distributed. It does not identify which spins will win, it does not suggest a betting pattern that reduces the house edge, and it does not turn a negative-expectation game into a positive one.
Progression systems illustrate the limit of the tool. A player who doubles an even-money bet after each loss changes the size of each wager, which changes the per-spin standard deviation in currency terms and concentrates risk into a small number of large bets. The expected loss per unit wagered is unchanged, because every unit still faces the same house edge. What the progression does is redistribute outcomes: many small wins become more likely, and a few very large losses become possible. The arithmetic of expected value and variance describes that redistribution; it does not remove it.
For a player who wants the number for a specific bet and wheel layout, the calculation is always the same three steps. Find the winning probability from the wheel's pocket count, find the payout from the table layout, then run the variance formula above and take the square root. The information needed is printed on the table or available in the game's information panel, and the wheel type determines the pocket count.
FAQ
Does standard deviation change with the size of my bet?
Yes, linearly. If the per-spin standard deviation for a one-unit bet is s, then for a bet of b units on the same outcome it is b × s. Doubling the stake doubles the standard deviation in currency terms, while the expected loss per unit wagered stays the same.
Can I use standard deviation to choose a roulette bet?
You can use it to understand the shape of the risk you are taking. Two bets can have the same expected loss per unit but very different standard deviations, so the same bankroll will survive very different numbers of spins. The choice is about how much swing you are prepared to sit through, not about changing the long-run edge.
Why does the American wheel change the calculation?
Because the winning probability changes. A single-number bet on an American wheel has a 1 in 38 chance rather than 1 in 37, and the payout remains 35 to 1. That changes both the expected value and the variance, and it increases the expected loss per unit wagered on every bet that pays at the same odds.
Gambling involves risk. Please only gamble with funds that you can comfortably afford to lose. While we do our best to provide good quality information, we cannot be held responsible for any loss you may incur from gambling. We do not provide legal, financial, or tax advice. Always check your local laws and the operator's terms before playing. If you or someone you know is struggling with gambling, help is available through responsible gambling resources. Please play responsibly.
Play responsibly
Gambling should be treated as paid entertainment, never as a way to earn income or recover losses. 1-800-MY-RESET (1-800-697-3738)
18+ or 21+ depending on where you are; follow the minimum age that applies to you.
This article is general information about how these mechanics work. It is not legal advice and not a recommendation to gamble or to use any particular operator. Availability and legality differ by jurisdiction — check the rules that apply where you are.