The Gambler's Fallacy: Why a String of Losses Doesn't Make a Win "Due"
Six reds in a row on a roulette wheel does not make black more likely on the seventh spin. This misconception, that a streak makes the opposite outcome "due," is the gambler's fallacy, and it survives because it feels like reasoning even though the underlying process it describes has no memory at all.
The misconception, stated plainly
The gambler's fallacy is the belief that if an outcome hasn't occurred for a while, it becomes more likely to occur soon, to "balance out" a run in the other direction. It shows up as "black is due," "the machine hasn't paid in ages so it's about to," and "three heads in a row, next one's definitely tails." All three statements assume the process is tracking its own history and adjusting future outcomes to compensate. Independent random processes do not do this.
The maths: why independence rules this out
A process is independent when the outcome of one trial has no effect on the probability of the next. A fair coin, a roulette wheel, and an RNG-driven slot spin are all independent processes by design and by regulation. Independence means the probability of heads on any single flip is 0.5, full stop, regardless of what the previous 10, 100, or 10,000 flips did. The coin does not know its own history. Neither does the wheel. Neither does the RNG.
Worked calculation: the actual probability of a streak
Here is the calculation that gets misapplied. The probability of 6 heads in a row, before you start flipping, is 0.5 raised to the 6th power: 0.5 x 0.5 x 0.5 x 0.5 x 0.5 x 0.5, which equals 0.015625, or roughly 1.6%. That's a genuinely low probability, and it's the correct one to quote before the sequence happens.
But once you're standing at flip 7, having already observed 6 heads, the question has changed. You are no longer asking "what's the probability of 6 heads then a 7th outcome." You're asking "what's the probability of heads on this one flip," and for an independent process, that answer is 0.5, exactly what it was before the streak started. The low probability belonged to the whole sequence viewed in advance. It says nothing about the next single flip viewed after the fact, because the coin was never carrying the streak's improbability forward as some kind of pressure to correct.
Table: intuition versus actual probability
| Situation | What intuition often says | Actual probability |
|---|---|---|
| Coin lands heads 6 times in a row, before it happens | Roughly 1.6% likely | 1.6% (0.5^6), correct |
| 7th flip, after 6 heads already observed | "Tails is due," feels more likely than 50% | 50.0%, unchanged |
| Roulette lands red 8 times in a row, before it happens | Roughly 0.2% likely (approximating a fair wheel) | Roughly 0.2%, correct |
| Next spin, after 8 reds already observed | "Black is due" | Same probability as any other spin, unchanged by the streak |
| A slot hasn't paid a bonus in 200 spins | "It's about to hit" | Unchanged by spin count; each spin draws independently against the same fixed probability |
The pattern in this table is consistent: correct intuition and correct maths agree completely about a streak's probability before it happens, and they diverge sharply about what that streak implies afterward. The fallacy is specifically a reasoning error about the second column, not the first.
Where the confusion actually comes from
The gambler's fallacy isn't a failure to understand rare events, most people correctly sense that 6 heads in a row is unusual. The error is applying that correct rarity judgment to the wrong question. Before a sequence, "6 heads in a row" is one specific, unlikely outcome out of many possible sequences. After 6 heads have already landed, you're not asking about that sequence anymore, you're asking about one new, independent event, and the sequence that already happened has no causal channel through which to influence it. Confusing "this specific long sequence was unlikely" with "therefore the next single outcome is pulled toward correction" is the entire fallacy in one sentence.
A real wheel, not just a fair coin
European roulette is a cleaner illustration than a coin because its probabilities aren't a round 50%. A single-zero wheel has 37 pockets, 18 red, 18 black, 1 green zero, so the probability of red on any spin is 18 divided by 37, which equals 0.4865, or 48.65%, not 50%, and not 48.65% minus something for however long red has or hasn't shown. That figure is fixed by the physical wheel and stays fixed on every spin regardless of history. Suppose red hasn't landed in 8 spins. The gambler's fallacy says red is "overdue" and therefore more likely than 48.65% on spin 9. The wheel has no record of those 8 spins. Spin 9's probability of red is 48.65%, the same number it always is, because the ball and wheel carry no memory of prior outcomes into the next spin.
This is also why "the house edge will catch up" and "red is due" are not the same claim, even though they get confused for each other. The house edge, roughly 2.7% on a single-zero wheel, comes entirely from the green zero paying nobody on an even-money bet. It is baked into every single spin equally, from the very first spin of the day to the millionth. It does not need streaks, corrections, or overdue outcomes to operate, and it does not accelerate or soften based on what has or hasn't hit recently. Believing a streak changes the house edge misunderstands where that edge actually comes from.
Not to be confused with the law of large numbers
The law of large numbers is a real, correct statistical result, and it's worth separating from the fallacy precisely because the two get blended together. The law of large numbers says that as the number of trials grows very large, the observed average gets closer to the true probability. It says nothing about individual trials correcting each other to make that happen. A coin flipped a million times will land close to 500,000 heads not because any specific flip compensated for an earlier one, but because in a large enough sample, any short streak becomes statistically negligible against the total volume. The convergence happens through dilution, an enormous number of additional independent flips added to the pile, not through correction, some flips owing a debt to balance out others. Mistaking dilution for correction is most of what makes the gambler's fallacy feel like sound statistics when it isn't.
The mirror image: the hot hand belief
The reverse belief shows up too, that a streak in one direction predicts more of the same, "it's on a heater, ride it." This is sometimes called the hot hand fallacy in other contexts, and in an independent-outcome game it fails for the identical reason the gambler's fallacy does. Whether someone reads a streak as a signal to correct or a signal to continue, both readings assume the process is responding to its own history. An independent process is doing neither. It is simply drawing again from the same fixed probability it always draws from.
Why this matters specifically for casino games
Every regulated RNG-driven casino game is built and certified to produce independent outcomes, spin to spin and hand to hand. There is no mechanism, hidden or otherwise, tracking how long it's been since a specific outcome occurred and adjusting the next draw to compensate. A machine or a table "being due" describes a belief about the game's internal state that the actual system does not have. The RTP figure attached to a game is a long-run average calculated across an enormous number of independent trials, and that average has no bearing on, and receives no correction from, any individual trial or short streak within it.
What this means at the table
A losing or winning streak changes your bankroll. It does not change the probability of the next outcome, and no amount of tracking a streak, on paper or in your head, turns it into information about what comes next. The only number that reliably describes your position after a streak is your remaining balance, not an adjusted probability for the next bet.
Responsible Gambling
This piece describes how independent probability works, not a system for predicting or improving outcomes, and no observed streak changes the odds of what happens next. Confirm any UK operator holds a UK Gambling Commission licence before playing. GAMSTOP offers free self-exclusion across every licensed UK operator, and BeGambleAware provides confidential support for anyone who wants it.