The idea in one paragraph
The law of large numbers says that as you repeat a chance process more and more times, the average of the results tends to move closer to the true expected average. Flip a fair coin enough times and the share of heads will typically end up very near 50%. The law is real and important, and it is also widely misunderstood, especially about how fast it works and what it does not promise.
A word on context: slots are games of chance with a built-in house edge. This law explains why that edge matters over time, and it does not mean any player can expect to come out ahead. Most players lose money over time, and nobody should gamble money they cannot afford to lose.
A coin example with numbers
For a fair coin, the true chance of heads is 0.5. Here is what typically happens as flips accumulate (the figures are approximate and typical, not guaranteed):
| Number of flips | Typical share of heads | Typical gap between heads and tails counts |
|---|---|---|
| 10 | anywhere from about 35% to 65% | a few flips |
| 100 | about 45% to 55% | around 10 |
| 10,000 | about 49% to 51% | around 100 |
Look at the last column. The typical gap in counts grows from a few to around 100 as flips rise from 10 to 10,000. The share gets steadier because the gap grows far more slowly than the number of flips: 100 out of 10,000 is just 1%.
So the law works on averages and shares, not on raw counts. It does not say heads and tails will become equal in number.
What the law does not say
It does not say a deficit will be repaid
Suppose 10 flips give 7 heads. Many people feel that the next flips must favor tails to even this out. They do not. If the next 90 flips are perfectly typical, with 45 heads, the total is 52 heads out of 100, or 52%. The early excess remains in the record; it is simply diluted. This is the confusion behind the gamblers fallacy, and it is worth keeping the two ideas apart. Because each flip is among independent events, nothing steers the future.
It does not promise anything in a short run
The law says that results approach the average over very many trials. It gives no guarantee about 50, 500 or even 5,000 trials. A session of slot play is usually a short run by this standard, so the outcome can land far from the long-run average in either direction. This is the problem with small samples, and the reason one lucky or unlucky night says little about the true odds.
It does not say “it will be your turn”
The long-run average belongs to the whole process, not to any individual player’s account. No individual gets credit for time spent losing.
Why a casino sees steady results
The law explains a seeming paradox. A single player’s session can be a big win or a big loss, while an operator’s overall results over millions of bets are far more predictable. Each bet in a game that loses for the player on average carries a small average advantage for the house. With a huge number of independent bets, the actual results settle close to that average.
Take a hypothetical game with a 5% edge for the house, meaning the player’s expected loss is $5 per $100 wagered. Across $100 million in total wagers, the expected house result is about $5 million. The actual figure would usually be quite near that, relative to the amount wagered, because of the law of large numbers. For one player wagering $500 in a session, the result might be anything from a sizable gain to a loss of most of the stake.
This is the link between expected value and what happens in practice: the average is dependable in aggregate, and individual outcomes are not.
How slow is “very many”?
The spread in the average shrinks in proportion to the square root of the number of trials. That is slow. To make the typical miss ten times smaller, you need about one hundred times as many trials. For games with rare large payouts, where the spread of single results is wide, the number of trials needed before the average settles is far larger than most players ever experience. That’s one reason slot volatility matters: the more volatile the game, the longer it takes for the long-run average to show itself, and the wider the range of ordinary session results.
A published return to player figure is the same kind of long-run average. It is a statement about enormous numbers of spins, and it cannot tell you what your own session will return.
See it for yourself
A spreadsheet can show the law at work. Simulate thousands of coin flips, chart the running share of heads, and watch it wobble early and settle later. Then chart the running difference between heads and tails and see how it wanders rather than returning to zero. A step-by-step approach is described under chance simulations.
Key points
- Averages settle toward the true value as trials increase, but the raw gap in counts typically grows.
- Nothing “corrects” earlier results; they get diluted.
- The law gives no guarantee for a short session, and short sessions are what players experience.
- A game that loses on average for players can be reliably profitable for the house and still allow an individual player to be ahead for a while.
Because that last point can tempt people to keep going after a lucky spell, it helps to decide on a limit before playing and to stop when you reach it. Treat gambling as paid entertainment, never chase losses, and if it becomes a worry, reach out to a national problem-gambling helpline or local health service.