The law of large numberswhy the house wins
Short runs are noisy, long runs are honest. Randomness rules the next flip; arithmetic rules the next million.
Before this
Next in Statistics
The normal distributionStatistics, stop 9 of 102 minThis opens up
Transcript
It follows the voice. Click a sentence to jump there.
Flip a coin ten times and count the heads. Seven. Is the coin rigged? Before you accuse it, flip ten more: four heads this time. Ten more: six. Another round: seven again. A fair coin lands seven out of ten all the time, because ten flips is a tiny sample — and tiny samples exaggerate luck.
So keep flipping, and after every flip write down the running fraction of heads so far. Across the first ten flips the line lurches wildly with every toss. By a hundred flips the swings have tightened. By ten thousand, the line has all but glued itself to one half — the coin's true rate.
That settling has a name: the law of large numbers. As the number of tries grows, the average of the results closes in on the true odds. Short runs are noisy — a lucky streak drags the average around. Long runs are honest — hot streaks and cold streaks cancel, and what remains is the coin itself: one half.
Now run a casino. Rig a game gently: the house wins fifty-one percent of the time, the player forty-nine. On a single bet anything can happen — the house loses that one almost half the time. But the house never plays once. Over a thousand bets, the wobble around fifty-one shrinks. Over a million, it nearly vanishes, and the two-percent edge becomes steady, near-certain profit. The house doesn't need luck. It needs volume.
The territory splits in two. Randomness owns the next flip: no formula can call a single toss, and on any one night a gambler can walk away rich. Arithmetic owns the next million: enough bets pin the average to the odds. Seven heads in ten was noise, not evidence. The house never bets on a flip — it bets on the average, and the average keeps its promises.