From a few hands to thousands
Over a short session, luck dominates, and results are close to a coin flip. As the number of hands grows, the picture changes in a predictable way: the average result per hand settles towards the expected value, which for a blackjack player is a small loss per unit bet. This article follows that process over thousands of hands and explains why a good or bad stretch tells you very little about what comes next.
The figures are illustrations from a simple model, and they are not measurements of any specific table.
The model, briefly
Assume flat one-unit bets, basic strategy, an average loss of about 0.5% of each unit wagered, and a standard deviation of about 1.1 units per hand. For N hands:
- Expected result: minus 0.005 x N units
- Standard deviation of the total: about 1.1 x the square root of N units
The expected loss grows in step with N, while the spread grows only with the square root of N. That difference is the whole story: given enough hands, the steady drift beats the random swings.
What the numbers look like
| Hands | Expected result | Standard deviation | Rough chance of finishing ahead |
|---|---|---|---|
| 1,000 | minus 5 units | about 35 units | about 44% |
| 10,000 | minus 50 units | about 110 units | about 33% |
| 100,000 | minus 500 units | about 348 units | under 10% (about 7.5%) |
Check the 10,000-hand row: 0.005 x 10,000 = 50, and 1.1 x 100 = 110. The expected loss is 50/110, or about 0.45, standard deviations below zero. A bell-shaped approximation puts roughly a third of outcomes above zero. At 100,000 hands, 1.1 x the square root of 100,000 is about 348, and the expected loss of 500 is about 1.4 standard deviations away, which leaves only a few percent of outcomes ahead. By a million hands the expected loss is roughly 4.5 standard deviations away, and finishing ahead is for practical purposes not going to happen.
Notice how slowly this plays out. Over 1,000 hands the player is still not far from an even chance of being ahead. This slow convergence is one reason the house edge can be hard to see from personal experience.
The average per hand converges
Another way to see the same thing is to look at average result per hand. The spread of that average is about 1.1 divided by the square root of N:
- After 10,000 hands: about 0.011 units per hand, or 1.1%, which is more than twice the 0.5% edge. The observed average can easily look like a profit or a much larger loss.
- After 100,000 hands: about 0.0035 units per hand, or 0.35%.
- After 1,000,000 hands: about 0.0011 units per hand, or 0.11%.
Only with very large samples does the observed average settle close to the true figure. This is the law of large numbers, and it applies equally to personal records, casino records and simulations. It is also why researchers rely on Monte Carlo methods and simulation basics to estimate odds, running millions of hands to get stable values. The uncertainty in an estimate from a limited number of hands can be expressed with confidence intervals.
Why short-term luck does not predict the long run
If you are ahead after 500 hands, the future 10,000 hands still start from the same odds. The earlier gain does not vanish or grow; it just becomes a smaller share of the result as more hands are added. What changes is that the expected loss over the future hands is added to wherever you stand. The same applies in reverse: an early loss is not “owed back”.
This matters for several common beliefs:
- A winning streak is not evidence of skill, a lucky seat, or proof that hot tables exist. Short samples regularly produce impressive-looking runs.
- A losing streak is not evidence of a rigged game. Long runs are routine, as the numbers on losing streaks show.
- A good week does not predict a good month.
Many popular blackjack myths grow from drawing conclusions from small samples.
What changes the long-run figure
The expected loss per hand depends on:
- the house edge of the specific rule set, including natural payouts and dealer rules;
- how closely the player follows basic strategy;
- how much is wagered in total, which depends on bet size and the pace of play.
Better play reduces the loss rate without reversing it. Worse play, or a worse table, increases it. Because the total expected loss is the loss rate multiplied by the amount wagered, the same rate at higher stakes or faster pace costs more. The cost can be estimated from hands per hour and average bet, though the estimate is only as good as its inputs.
What this means for a player
There is a practical, non-mathematical side to all this. Results that look like success after a few hundred hands are not likely to last through thousands of hands, so judging a method by recent results is unreliable. The numbers also argue for deciding in advance how much entertainment is worth paying for, and stopping at that point. Blackjack is a negative-expectation game, and nothing in strategy, timing or bet size reverses that.
If gambling stops being enjoyable or starts to cause problems, a local support service is a good place to start.