A Super Mega Bowl guide
Flip a fair coin five times and get five heads. What are the odds the next flip is tails? If your gut says “high — it’s due,” you’ve just met the gambler’s fallacy: the mistaken belief that past independent outcomes change future ones.
A fair coin flip is an independent event. The coin has no memory, no ledger, no obligation to balance the books. The probability of heads is 0.5 on the first flip and 0.5 on the thousandth, no matter what came before. Five heads in a row is genuinely unlikely in advance — (½)⁵ ≈ 3.1% — but once those five have happened, they’re history. The sixth flip is a brand-new trial: still 50/50.
The fallacy conflates two different questions:
The streak feels like it’s “loading the dice” toward tails. It isn’t. That feeling is the fallacy.
The most famous real-world case played out at Monte Carlo in 1913: a roulette wheel came up black twenty-six spins in a row. As the run stretched on, players grew convinced the wheel was “overdue” for red and kept piling money against black — and lost a collective fortune. Nothing was overdue. On a single-zero wheel, black lands with probability 18/37, so a run of 26 blacks had only about a 1-in-137-million chance measured before it began — yet on every individual spin during the streak, red and black stayed exactly equally likely. The imbalance the gamblers thought they saw existed only in their heads.
Humans are pattern-matchers who intuitively expect small samples to mirror the long-run average — a bias Kahneman and Tversky called belief in the “law of small numbers.” The real law of large numbers says the proportion of heads converges to 0.5 as the number of flips grows toward infinity. Our intuition wrongly compresses that into “it should balance out soon,” over the next few flips. It doesn’t. The average is reached not by tails “catching up,” but by early streaks being diluted by an ever-larger pool of later flips.
Flip the fallacy over and you get the opposite error — assuming a streak will continue (“this slot machine is hot”). Both mistakes share one root: reading meaning into independent randomness. The correct lens is regression to the mean — extreme runs are usually followed by more ordinary results, not because of a corrective force, but simply because extremes are rare and average outcomes are common.
The fastest cure is data. Flip a large batch and watch what actually happens:
No independent event is ever “due.” Not a coin, not a die, not a roulette wheel, not a lottery number. The past doesn’t owe the future a correction. Internalize that and you’ll out-reason every gambler chasing a pattern that was never there.