The belief that luck must balance
After five heads in a row, many people feel that tails is more likely on the next flip. The intuition is that a fair coin should produce about half heads and half tails, so a surplus of heads must be corrected soon. This is the gambler’s fallacy: the belief that a run of one outcome in independent random trials makes a different outcome more likely next.
For a fair coin, the chance of heads on the next flip is 1 in 2 no matter what came before. The coin does not know its history. The same is true of a fair die, a well-shuffled deck for each fresh shuffle, or any process where each trial is unaffected by the others.
Why the intuition feels reasonable
It rests on a real fact misapplied. Over many flips, the proportion of heads tends to settle near one half. That is the law of large numbers. But it does not work by compensating for earlier results. It works by dilution: later flips carry on at 50-50 and gradually swamp the early surplus.
Here is a quick illustration. Suppose that after 10 flips you have 8 heads, a surplus of 3 over the expected 5. If the next 90 flips split evenly, 45 and 45, you will have 53 heads in 100, which is 53 percent, close to one half. The imbalance was never “corrected”. It just became a smaller share of a bigger total. The expected count of extra heads stays at 3, while the proportion drifts toward 50 percent.
People also tend to judge small sequences as if they should look like the long-run average. They expect a short random sequence to be “balanced” and to alternate often, and they find real randomness suspicious when it clusters.
Streaks are normal
Random sequences contain longer runs than most people expect. The chance of getting heads on five specific flips in a row is (1/2)⁵ = 1/32, about 3.1 percent. That sounds rare, but in a long series of flips there are many places a run of five could begin. In 100 flips, a longest run of six or seven in a row is quite typical, while a sequence that never exceeds three or four would be unusually regular.
A simple test: ask a friend to write down a “random” sequence of 100 flips from their head, while you flip a coin 100 times. The human version tends to have shorter runs, because people avoid long ones. The real coin’s list often has a streak that looks designed.
When the fallacy bites in games
Dice and cards. A player who has not rolled a particular number in a while may feel it is overdue. The die is unchanged. In a card game with a fresh shuffle each hand, a long stretch without a strong hand does not raise the chance of one next time.
Loot and drops. Failing many times to get a rare item does not make the next attempt more likely to succeed in a plain random system. Whether it does depends on the rules: some games use pity timer or pseudo-random distribution systems to adjust the odds as you fail, which are deliberate design choices and not natural features of chance. Knowing which kind of system you are in matters. If you do not know the mechanism, “due” is a guess.
Roulette and similar games. At the Monte Carlo casino in 1913, the ball is widely reported to have landed on black 26 times in a row, and gamblers lost heavily betting on red. Whatever the exact details, the lesson is the same: each spin was independent, so the long run on black did not make red more likely. Since the casino keeps a built-in house edge, expecting a bet to be “due” gives no advantage.
Care with the reverse mistake
Sometimes a streak is evidence of something. If a coin lands heads 30 times in a row, the sensible reaction is to doubt that it is fair, not to bet on tails. The independence assumption can fail: a die might be loaded, a game might use a non-uniform system, or a player’s skill might be changing. This is a different matter from the gambler’s fallacy, which applies when independence and fairness do hold.
The conditional logic runs both ways. Under a fair-coin assumption, the past tells you nothing about the next flip. But the past can tell you about the fairness assumption itself. A short streak barely moves your belief, because short streaks are common under fairness. A very long one moves it a lot. If you want a formal way to check, a chi-square test can compare observed counts with what fairness predicts.
Practical habits
- Ask whether trials are independent. If the rules say each result is unaffected by prior ones, the past cannot help predict the next.
- Check whether you are in a system with memory. Drawing without replacement is one: when cards leave a deck, the remaining odds do change. That is real dependence and a different matter from a fair coin.
- Do not size decisions by “due”. Feelings of being due should not influence how much you commit.
- Look at long records. Judge frequencies over many trials, not from the last few.
- Remember that outcomes are uncertain. No pattern in a random process guarantees what comes next, and no betting plan based on streaks changes the odds.
The fallacy is intuitive because balance is a pleasing idea. Random processes simply don’t care about it. Understanding that is one of the more useful steps for thinking clearly about chance.