The Law of Large Numbers
The law of large numbers is the quiet engine behind every casino on earth. It's a simple, ironclad piece of mathematics: the more times you repeat a random trial, the closer the observed average creeps to the true expected value. It's also the law that people most often confuse with its evil twin — the gambler's fallacy. Getting the two straight is the difference between understanding gambling and being fooled by it.
What the law actually says
Flip a fair coin once and you get either 0% or 100% heads — nowhere near the "true" 50%. Flip it ten times and you might get 7 heads, or 70%. Flip it ten thousand times and you'll land remarkably close to 50%. The law of large numbers says that as the number of trials grows, the average result converges to the expected value. Not eventually balances out — converges. The proportion gets closer and closer to the true probability, and stays there.
| Coin flips | Heads (typical) | Observed % heads |
|---|---|---|
| 10 | 7 | 70% |
| 100 | 54 | 54% |
| 1,000 | 517 | 51.7% |
| 100,000 | 50,140 | 50.14% |
Look at the last column. The percentage marches steadily toward 50%. But here's a subtlety most people miss: the raw count can drift further from "perfectly even" even as the percentage tightens. After 100,000 flips you might be 140 heads "ahead" — a bigger gap than after 10 flips — yet as a fraction it's now a rounding error. The average converges; the running total need not.
Why this is NOT the "law of averages"
The gambler's fallacy — sometimes called the misnamed "law of averages" — claims that if a coin has come up heads five times in a row, tails is now "due." This is flatly wrong. The coin has no memory. The chance of tails on the next flip is still exactly 50%, regardless of history. Past results do not reach forward and correct the future.
So how does the average ever converge if the universe never "evens things out"? The answer is beautiful: it converges by dilution, not correction. That early run of five heads doesn't get cancelled by a future run of tails — it just gets swamped by the sheer volume of later flips, each sitting near 50%. The imbalance isn't erased; it's drowned out.
Key idea: the law of large numbers does not say a streak will reverse. It says a streak becomes insignificant once you average it against a flood of later, independent trials. Nothing is ever "due."
Why the house always wins over volume
Now apply this to a casino. Every game is built with a small negative expected value for the player — the house edge. Suppose a game has a 2% edge, meaning the expected return is $0.98 per $1 wagered. For a single player on a single night, that 2% is almost invisible; luck dominates, and plenty of players go home winners.
But the casino isn't a single player. It books millions of bets a day across thousands of players. At that scale, the law of large numbers takes over completely: the casino's actual return locks onto that 2% edge with near-perfect reliability. Individual sessions stay wild and unpredictable — that unpredictability is exactly what keeps the games fun — but the aggregate is as steady as a utility bill.
This is the deep asymmetry of gambling. The player experiences the small-sample world, where anything can happen in a night. The house lives in the large-sample world, where the edge is destiny. The same math that makes your evening exciting makes the operator's revenue boring and certain.
The takeaway for any player
No betting system can beat a negative expected value, because no system can change the law of large numbers. Doubling after losses, chasing streaks, switching tables — these just rearrange the order of bets. The more you play, the more tightly your results hug the house's edge. The only way to avoid the long-run grind is to not be in it for the long run at all.
On Riskr the money is fake, so you can play thousands of rounds and watch the law of large numbers pull your results toward the math — without it costing you a thing. The only stake is your spot on the leaderboard.