The Gambler's Fallacy & Hot Hand: Empirical Data from Casinos
Real gamblers, real money, real casino footage — and a real, measurable bias.
Lab experiments have shown for decades that people misjudge random sequences. This study asked a sharper question: does that bias actually show up when real gamblers are betting their own money on a real casino floor?
Study overview
Two economists, Rachel Croson and James Sundali, wanted to test two well-known cognitive biases outside the psychology lab: the gambler's fallacy (expecting a color to be "due" after a streak of the opposite color) and the hot hand fallacy (expecting a streak to continue and betting bigger because of it). Both biases assume patterns exist in sequences that are actually independent — each spin of a fair wheel has no memory of the last one.
Methodology
Instead of running a lab experiment with student volunteers, the researchers obtained security camera footage from a casino in Reno, Nevada: 18 hours of live roulette play, recorded in three separate six-hour blocks over three days in July 1998. The camera gave an overhead view of the betting layout, letting the researchers track individual bets by chip color and position without identifying who was playing. They then compared how betting on red or black shifted after real winning and losing streaks against what a bettor with no bias at all would be expected to do.
Key findings
The researchers found small but statistically significant evidence of both biases among real casino patrons. Players were somewhat more likely to bet against a color after it had come up several times in a row — the gambler's fallacy in action — and separately showed hot-hand-style behavior, increasing their own bet sizes after personal winning streaks. Both effects were present in the same population of real money bettors, not just in controlled lab settings.
Limitations
The authors themselves describe the biases they found as small — this is evidence that a measurable effect exists, not that most players are dramatically irrational most of the time. The data comes from a single casino, a single table, and a three-day window in 1998, so results may not generalize perfectly to every casino, era, or player population. The video-based method also couldn't capture individual player characteristics — experience level, for instance — that might explain who is more or less prone to these biases.
Practical meaning
This study is genuine evidence that the gambler's fallacy isn't just a psychology-101 curiosity — it shows up in how people actually bet their own money. The practical takeaway is simple and doesn't require exaggerating the study's findings: however many times red has come up in a row, black is not "due," and treating it as more likely is exactly the bias this research documents. The same goes in reverse for chasing a perceived hot streak with bigger bets. Following a predetermined staking plan — like the ones described in the systems library — removes this particular temptation from the decision, whatever you think of the systems themselves.