The Gambler's Fallacy: Why Past Results Don't Predict the Future
One of the most persistent and costly cognitive biases in games of chance—and how to recognize it in yourself.
What Is the Gambler's Fallacy?
The gambler's fallacy is the mistaken belief that if something happens more frequently than normal during a given period, it will happen less frequently in the future (or vice versa). It's the feeling that after five heads in a row, tails is "due"—that the universe somehow keeps track and will balance things out.
Here's the truth: the coin doesn't know what happened before. Each flip is independent. The probability of heads on the sixth flip is exactly 50%, regardless of whether the previous five were all heads, all tails, or any other combination.
The Monte Carlo Fallacy
This bias is sometimes called the "Monte Carlo fallacy" after a famous 1913 incident at a Monaco casino. The roulette ball landed on black 26 times in a row. As the streak continued, gamblers bet increasingly large amounts on red, convinced it was due. The casino made millions as black kept coming up. Those bettors fell victim to the exact same probability every spin: roughly 48.6% for red.
Why We Fall for It
The gambler's fallacy stems from a deep human tendency to see patterns and expect balance. Our brains evolved to find meaning in sequences—it helped our ancestors detect predators, predict weather, and remember where food was. But randomness doesn't follow our intuitive expectations about fairness and equilibrium.
We also tend to misunderstand what "random" actually looks like. A sequence like H-T-H-T-H-T feels more random than H-H-H-T-T-T, even though both have the same probability. True randomness often has streaks and clusters that our pattern-seeking minds interpret as meaningful.
Real Examples in Action
Lottery Numbers
"Number 23 hasn't come up in months—it's due!" No, it isn't. Lottery drawings are independent events. A number that hasn't appeared recently has exactly the same probability as one that appeared last week. Some lottery websites even show "hot" and "cold" numbers, implying patterns that don't exist in truly random drawings.
Roulette and Casino Games
Casinos profit from the gambler's fallacy. They display boards showing recent roulette results, encouraging players to bet on "due" numbers. But the wheel has no memory. Every spin is independent of every other spin, and the house edge remains constant.
Sports Betting
"This team has lost three games in a row—they're bound to win the next one!" Teams aren't coins; there may be real factors affecting performance. But the mere fact of previous losses doesn't increase the probability of a win. Regression to the mean is real, but it's not magic—it's just statistics working over large samples.
Birth Rates
A family with four boys might believe they're "due" for a girl. But each pregnancy is an independent event with roughly 50/50 odds (actually slightly favoring boys). Having had boys doesn't change the biological probabilities of the next child.
The Mathematics
Let's prove this with a simple example. What's the probability of getting heads on a fair coin flip?
- P(Heads) = 0.5 or 50%
What's the probability of getting heads after five consecutive heads?
- P(Heads | Previous 5 were Heads) = 0.5 or 50%
The conditional probability is the same because the events are independent. Independence means that knowing the outcome of previous events provides no information about future events. Mathematically:
The probability of getting 6 heads in a row is very low (1/64 or about 1.6%). But given that you've already gotten 5 heads, the probability of the 6th being heads is just 1/2. The previous flips are already done—they're not part of the probability calculation anymore.
Related Concepts
The Hot Hand Fallacy
The opposite of the gambler's fallacy: believing that a streak indicates continued success. "She made five shots in a row—she's on fire!" While skill can certainly create streaks in games of skill, in pure games of chance, a winning streak doesn't increase the probability of the next win.
Regression to the Mean
This is a real statistical phenomenon, but often misunderstood. Extreme results are typically followed by more average results—not because of any balancing force, but because extreme values are simply less common. It's not the universe correcting itself; it's just how probability distributions work over time.
How to Protect Yourself
- Remember independence: Ask yourself, "Does the outcome of this event actually depend on previous outcomes?" If not, previous results are irrelevant.
- Ignore streaks in random events: A coin that landed heads 10 times is still a fair coin with 50/50 odds.
- Be skeptical of "due" thinking: Nothing in probability is ever "due" in independent events.
- Understand the law of large numbers: Averages converge to expected values over many trials, but not through any compensating mechanism—just through new random data overwhelming old data.
- Use tools wisely: When you use random number generators, you're getting fresh, independent random values every time—no history, no bias.
Frequently Asked Questions
But eventually things have to even out, right?
Over a very large number of trials, the proportion of outcomes will approach expected values. But this happens because new data dilutes past imbalances—not because future outcomes "correct" for the past. After 10 heads in a row, you might eventually see 1000 heads and 990 tails. The proportion (50.5%) is close to 50%, but you still have more heads—the past was never erased.
What if a streak is really unusual?
Unusual things happen. A 1-in-a-million event happens to someone every day when millions of people are playing. A long streak might make you question whether the game is fair (is this coin actually biased?), but if the game is fair, past results still don't affect future probabilities.
Is this different for skilled games?
Yes. In games involving skill (like basketball or poker), past performance can indicate something real about ability, fatigue, or psychological state. But for pure chance—coin flips, lottery drawings, fair dice—each event is independent regardless of context.
Continue Learning
Generate Fresh Random Numbers
Every generation is independent—no memory, no patterns, just genuine randomness.
Open Lucky Numbers →