A single spin of a roulette wheel is basically a coin toss, yet stack up a few million spins and a casino’s profit margin becomes almost as predictable as gravity. That gap between chaos and certainty comes down to one law it took a mathematician twenty years to prove.
What the law of large numbers actually says
The law of large numbers says the average of many independent random results drifts closer and closer to the true theoretical value, the more times an experiment gets repeated.
Roll a single die enough times, and the running average of every roll creeps toward 3.5, the exact mathematical expectation of a six sided die. After a handful of rolls, results bounce all over the place. After thousands, they barely move from that number.
The same idea explains the casino. On any single spin of roulette, the house genuinely cannot predict who wins. Over tens of thousands of spins, its edge turns into a number precise enough to build a business plan around.
The coin flip myth that refuses to die
Here is where popular intuition gets it backwards. Plenty of people assume a long streak of heads should soon be balanced out by a run of tails.
It never works that way. Flip a coin a million times, and the ratio of heads to tails does settle close to fifty-fifty. But the raw, absolute gap between the total number of heads and the total number of tails keeps growing the entire time, just far more slowly than the number of flips itself.
That mismatch, a shrinking ratio next to a growing gap, is exactly what the law of large numbers predicts, and exactly what the gambler’s fallacy gets wrong.
It took one mathematician twenty years to prove it
Gerolamo Cardano floated the basic idea back in the 1500s, noticing that statistics seemed to sharpen as more data came in, but he never actually proved it.
That task fell to Jacob Bernoulli, a Swiss mathematician who spent more than two decades chasing a rigorous proof for simple win or lose outcomes. He finally published it in 1713 in a book called Ars Conjectandi, and gave his own result a proud nickname: his “golden theorem.”
More than a century later, in 1837, French mathematician Simeon Denis Poisson gave the law its now familiar name and added a proof of his own, split into two parts. Andrei Kolmogorov pushed the idea even further in the 1930s, nailing down the stronger version mathematicians still rely on today.
From the roulette table to a cloud of gas
The reach of this law goes well beyond gambling. A single molecule bouncing around a room behaves in a way that looks completely random and impossible to predict.
Put billions of molecules together, though, and their collective spread becomes orderly enough to be described with precise physics equations, the same ones used to explain how gas diffuses through air.
A casino’s balance sheet and a drifting cloud of gas end up governed by the same underlying mathematics: individual chaos settling into collective order.
Where the law quietly stops working
The law of large numbers is not a universal fix. It breaks down completely for so called heavy tailed distributions, like the Cauchy distribution, which has no defined average value to converge toward in the first place.
It also does nothing to erase selection bias. However many times a flawed experiment gets repeated, a skewed sampling method stays skewed. Averaging more results never fixes a broken starting point.
A pattern hiding inside every kind of chaos
Insurance companies, casinos, and physicists all lean on the very same centuries old idea to turn uncertainty into something they can actually plan around. The next time a single unpredictable event feels like pure chaos, it helps to remember that zoom out far enough, and even chaos tends to settle into a pattern.

