Gambler’s Fallacy: Why Past Spins Don’t Change the Odds

The gambler’s fallacy is the mistaken belief that a run of past results changes the probability of the next one. After five reds in a row on a roulette wheel, the chance of red on the next spin is exactly what it was before: 18/37, or about 48.65%, on a European wheel. The wheel has no memory, so nothing is “due.”

This matters because the fallacy feels reasonable. A long streak seems unusual, and it is: five reds in a row happens only about 2.72% of the time. But unusual before the streak starts is not the same as unusual after it has happened. Once the first five spins are history, only the next spin is still random, and its odds have not moved.

This guide defines the gambler’s fallacy, shows the formula that explains it, works through roulette, coin and dice examples with every step, and lists the habits of thought that keep the mistake alive. It also covers what the law of large numbers does and does not say, and how to avoid the fallacy in your own reasoning. Every figure is recomputed from the formulas shown, and all examples are hypothetical.

What Is the Gambler’s Fallacy?

The gambler’s fallacy (also called the Monte Carlo fallacy) is the belief that an independent random event becomes more or less likely because of what happened just before it. Two events are independent when the result of one gives no information about the other. Every spin of a fair roulette wheel, every flip of a fair coin and every roll of fair dice is independent of the one before.

The fallacy comes in two directions. The first says a streak must end soon: “it has landed red six times, so black is due.” The second, sometimes called the hot-hand belief, says a streak will keep going: “red keeps hitting, so red is hot.” Both treat past outcomes as if they changed the equipment’s behavior. For a fair, independent process, neither is true.

A useful way to separate the ideas is to compare two different questions. The first is “How likely is a run of six reds?” That is a question about a whole sequence before any spin happens, and the answer is small. The second is “Given five reds already, how likely is a sixth?” That is a question about a single spin, and the answer is just the probability of red. The fallacy mixes the two up, carrying the rarity of the sequence over to the single spin.

The intuition behind the mistake is the belief that short sequences should look like the long-run average. People expect a fair wheel to alternate red and black frequently, so a streak feels like a violation that must be corrected. In reality, random sequences contain streaks far more often than most people expect, as the table later in this guide shows. Another name for this expectation is the “law of small numbers”: the false idea that a handful of results must mirror the long-run odds.

The Formula Behind Independence

The math that settles the question is the multiplication rule for independent events, together with the definition of conditional probability. For independent events A and B, the chance that both happen is the product of their separate chances.

where:

P(A and B) = probability that both events happen

P(A) = probability of the first event (for example, red on spin 1)

P(B) = probability of the second event (for example, red on spin 2)

Conditional probability asks how likely B is once A is known to have happened.

where:

P(B | A) = probability of B given that A happened

P(A and B) = probability that both happen

P(A) = probability that A happens

Put the two together. When A and B are independent, P(A and B) = P(A) × P(B), so dividing by P(A) leaves P(B | A) = P(B). Knowing that A happened changes nothing. That one line is the gambler’s fallacy refuted: the probability of the next result given the previous ones equals the probability of the next result on its own.

A third formula measures how rare a streak is before it starts. For a single-trial probability p, the chance of n successes in a row is:

where:

p = probability of the result on one trial (red on a European wheel: 18/37 ≈ 0.4865)

n = number of consecutive trials

The assumptions matter. These results hold for a fair device with fixed probabilities, such as a properly balanced wheel, a fair coin or fair dice, where every trial is independent. A biased or faulty device is a different problem, because a long run could then be evidence about the device itself. Fixed odds and independent trials are the baseline for all the examples here, and they are the standard setup in the wider topic of probability in gambling.

Gambler’s Fallacy Example: A Roulette Streak

Suppose a European wheel (37 pockets: 1 through 36 plus a single zero) has just landed on red five times in a row. A bettor reasons that black is overdue. Here is the math, one step at a time.

Step 1: P(red on one spin) = 18 ÷ 37 ≈ 0.4865, or about 48.65%.

Step 2: P(5 reds in a row) = (18/37)⁵ ≈ 0.0272, or about 2.72%. That is roughly 1 in 37.

Step 3: P(6 reds in a row) = (18/37)⁶ ≈ 0.0133, or about 1.33%.

Step 4: P(6th red | 5 reds already) = P(6 reds) ÷ P(5 reds) = (18/37)⁶ ÷ (18/37)⁵ = 18/37 ≈ 48.65%.

Step 5: P(black on the next spin) = 18/37 ≈ 48.65%, and P(green zero) = 1/37 ≈ 2.70%. These add to 100%.

The sixth spin is exactly as likely to be red as the first spin was. A bettor who doubles up on black because it is “due” is staking money on a probability that has not changed. The 2.72% figure describes the five-spin sequence before it happened; once those spins are done, they are fixed history and carry no force. The result also shows why this kind of reasoning is connected to the Martingale system, which depends on the belief that a streak of losses must end.

One more check from the same wheel: on an American wheel (38 pockets, two zero pockets), P(red) = 18/38 ≈ 47.37%, and five reds in a row has probability (18/38)⁵ ≈ 2.38%. The streak is a little less likely there, but again the next spin is just 18/38.

The Fallacy Across Coins, Dice and Streak Lengths

The same pattern appears in every game built on independent trials. The table below shows how rare streaks of red are on a European wheel, using P(n reds in a row) = (18/37)ⁿ. The “1 in” column shows roughly how many separate runs of n spins you would need for one to be all red.

Coins give the cleanest example. A fair coin lands heads with probability 0.5. Ten heads in a row has probability 0.5¹⁰ = 1/1,024 ≈ 0.098%. But after nine heads, the chance of a tenth is still 0.5, because P(10 heads) ÷ P(9 heads) = (1/1,024) ÷ (1/512) = 0.5.

Dice work the same way. With two fair dice, P(7) = 6/36 ≈ 16.67% on every roll. The chance of going ten rolls without a seven is (30/36)¹⁰ ≈ 16.15%, so such droughts are common. After nine rolls without a seven, the next roll is still a seven with probability 1/6. A number that has not appeared for a long time is not owed an appearance.

This is also why the gambler’s fallacy does not rescue any bet: edge and payout are fixed by the game, a point covered in the roulette house edge guide. A long dry spell never moves the house edge.

Law of Large Numbers and the Fallacy

The most common defense of the fallacy is the law of large numbers, which says that over many trials the observed frequency gets close to the true probability. It is true, but it does not say what people think it says.

The law works by dilution, not correction. Suppose a fair coin has landed heads 10 times in the first 10 flips. Over the next 990 flips, the expected number of heads is 990 × 0.5 = 495. The expected share of heads after 1,000 flips is (10 + 495) ÷ 1,000 = 0.505, or 50.5%. The early excess of heads is not cancelled by extra tails; it is simply outweighed by 990 ordinary flips. The expected excess stays at 5 heads, but as a percentage of the total it fades as the flips pile up.

This relationship between short-run swings and long-run averages is the subject of variance and standard deviation: the typical size of a swing grows with the square root of the number of bets, while the long-run percentage settles near the true probability. The law of large numbers describes where frequencies end up, never what the next trial will do.

Common Mistakes and Misconceptions

Several errors keep the fallacy alive. The first is confusing the odds of a sequence with the odds of the next trial. Five reds in a row is rare, but the sixth spin is a fresh 18/37.

The second is assuming streaks signal that something is about to change. Streaks of every length occur in random sequences, and the table above shows their normal frequency. The third is applying the fallacy in reverse: believing a streak will continue is the same error, because it also treats past results as informative.

The fourth is expecting the odds to balance within a session. Fair devices have no mechanism to even out short-run results. The fifth is thinking a bet at long odds is safer after a dry spell, which ignores that the odds are unchanged. A final caution: the reasoning applies only to independent trials. Drawing cards without replacement is different, because each card removed does change what remains. Gambling should be treated as paid entertainment, and none of this arithmetic is a way to gain an advantage over a negative-expectation game.

Quick Reference: Check Your Reasoning

Use this short test before trusting any streak-based reasoning. First, ask whether the trials are independent, such as spins, flips and rolls of fair equipment. Second, if they are, write down the single-trial probability and use it unchanged for the next trial. Third, use pⁿ only to describe how rare a streak was before it began.

To check your answer, confirm that the conditional probability equals the single-trial probability, and that the probabilities of a complete set of outcomes sum to 100% (for the European wheel: 18/37 + 18/37 + 1/37 = 37/37).

Related Concepts and Next Steps

The gambler’s fallacy sits at the center of several linked ideas. The broader rules for combining chances are covered in probability in gambling, including how to count outcomes and use the multiplication rule. For the most famous system built on the fallacy, see the Martingale system, which tests the “a win must come” belief with real streak and stake numbers. For how large swings can be while the odds stay fixed, read variance and standard deviation. To see what a bet costs in the long run, regardless of streaks, work through the roulette house edge calculation.