What standard deviation measures
Standard deviation is a measure of how widely results spread around their average. If every hand returned exactly the same amount, the standard deviation would be zero. In blackjack, a hand can lose, push or win, and sometimes pays more, so the result varies a lot from hand to hand.
The average result per hand is the expected value, and for blackjack it is a small negative number under normal rules. The standard deviation is a separate quantity: it describes how far an individual hand typically lands from that average.
A typical figure
For a one-unit starting bet, the standard deviation per hand in blackjack is usually quoted as a little over one unit, commonly around 1.1 to 1.15 depending on the rules and the strategy. This article uses 1.1 as a round illustrative figure.
Why is it above one? If every hand were a simple win or loss of exactly one unit, the standard deviation would be 1. Blackjack adds outcomes that stretch the range:
- doubling down puts two units at risk
- splitting creates two hands, each of which can win or lose
- a natural paying 3 to 2 returns 1.5 units
- pushes return zero, which pulls the spread slightly the other way
The net effect is a spread of a little more than one unit. The exact value changes with the rule set and how often the player doubles and splits.
How the spread scales
Hands are treated as independent, because each round is dealt from a freshly mixed or largely unchanged shoe. For independent results, variances add, and standard deviations add as the square root of the sum of squares. With identical hands, this gives a simple rule:
spread over n hands = standard deviation per hand x the square root of n
Using 1.1 per hand:
| Hands | Square root | Spread of total (units) |
|---|---|---|
| 1 | 1 | 1.1 |
| 100 | 10 | 11 |
| 400 | 20 | 22 |
| 10,000 | 100 | 110 |
Check: the square root of 400 is 20 and 1.1 x 20 = 22. The square root of 10,000 is 100 and 1.1 x 100 = 110.
Quadrupling the number of hands only doubles the spread. The spread grows, but more slowly than the number of hands.
Comparing the spread with the edge
The expected loss grows in direct proportion to the hands, while the spread grows with the square root. Take an illustrative edge of 0.5 percent against the player and one-unit bets:
- After 100 hands the expected loss is 0.5 units, against a spread of 11 units.
- After 10,000 hands the expected loss is 50 units, against a spread of 110 units.
- After 1,000,000 hands the expected loss is 5,000 units, against a spread of 1,100 units.
Only in the last case does the average loss clearly exceed the noise. For typical sessions of a few hundred hands, the results are dominated by variance, which is why a short session can end well ahead or well behind. This idea sits at the heart of statistical variance and is why a single session says little about a game’s true edge.
Using the figure for a rough range
A common rule of thumb is that about two-thirds of outcomes fall within one standard deviation of the average and about 95 percent within two. For 10,000 one-unit hands with 110 units of spread, a rough 95 percent range is the expected result plus or minus about 220 units. As a rough guide, this is an approximation using a bell-curve shape, and real distributions have heavier tails for rare large results.
Bet size matters
The spread is measured in units of the bet. With a 10-unit bet per hand, multiply all figures by 10: the spread over 100 hands would be 110 currency units, with expected loss scaling by the same factor. Bigger stakes magnify both the average cost and the swings.
Caveats
- The figure is approximate and rule-dependent.
- It assumes equal bets; varying bets, as in a bet spread, changes the calculation.
- It describes the distribution of results, not the likelihood of the next hand.
- Variance cannot be removed by strategy. Good play lowers the average cost, not the swings.
Blackjack has a house edge, and large swings occur in both directions. Set a limit you can afford, avoid chasing losses, and use local support services if gambling is causing harm.