What it means
The classic illustration is a fair coin that lands heads five times in a row. Many people feel that tails is now overdue, but the coin has no record of the previous flips, so the probability of tails on the sixth throw is still exactly one half.
The pull of the fallacy comes from a real statistical truth applied at the wrong scale. Over thousands of flips the ratio does settle near 50/50, but it gets there by swamping the early run with later results, not by producing compensating tails to balance the books.
In business the fallacy hides inside ordinary-sounding statements. A sales director who says the team has lost five pitches in a row so the next one must land, or an investor who buys a stock purely because it has fallen six days running, is making exactly the same error in a suit.
The antidote is to separate independent events from genuinely linked ones. Ask what the underlying probability is, whether anything about this attempt has actually changed, and whether a run of results is evidence of bad luck or evidence that the underlying probability was never what you assumed.
There is a real exception worth knowing, because not every sequence is independent. Drawing cards without replacement genuinely changes the odds for the next card, and some financial series show mean reversion driven by economics rather than by cosmic bookkeeping, so the skill lies in telling those cases apart.
The mirror image of the fallacy is just as costly. Where the gambler's fallacy says a streak must break, the hot-hand belief says a streak must continue, and both replace an honest estimate of the base rate with a story built from a very small sample.
In practice
Real-world examples.
Example
A retail investor watches a share price fall for seven consecutive sessions and buys heavily on the eighth, reasoning that a bounce is due. The price keeps falling because the decline reflects a deteriorating order book, not a random streak that owed anyone a rebound.
Example
An insurance underwriter has seen no large claims from a portfolio for three years and starts assuming the region is quieter than the model suggests. The next storm season produces two major claims, showing that the quiet run was ordinary variation rather than a change in underlying risk.
Example
A recruiter has made four hires in a row who left within a year and concludes the fifth is bound to work out. A colleague points out that the sensible response is to examine the screening process, since a run that long may signal a flawed method rather than bad luck.
Think of it
“Gambler's fallacy is thinking you're 'due' for a win-past randomness doesn't predict future.
Formula
Calculation
For independent events each with probability p, the chance of the same outcome n times in a row is p to the power of n. Crucially, the probability of the next event on its own remains p regardless of what came before.
Take a fair coin, where p = 0.5. The chance of six heads in a row is 0.5 x 0.5 x 0.5 x 0.5 x 0.5 x 0.5 = 1/64 = 1.5625%, which is about 1.6% and genuinely rare. But after five heads have already landed, the chance the sixth is also a head is simply 0.5, or 50%, and the sequence "five heads then a tail" also has a probability of 1/64, exactly as unlikely as six heads.
Now the sales version. A team with a 25% win rate faces pitches that are effectively independent, so the chance of losing five in a row is 0.75 x 0.75 x 0.75 x 0.75 x 0.75 = 0.2373, or 23.7%, which will happen roughly once in every four runs of five pitches. The sixth pitch still has a 25% chance of winning, not a higher one, and over 20 pitches the team should expect 20 x 0.25 = 5 wins whatever order they arrive in.Case study
Seen in the real world.
Calderwood Trading Partners is a fictional firm invented for this illustrative example. One of its junior traders adopted a position-sizing rule of his own devising: after each losing trade he doubled the size of the next one, on the reasoning that a winner had to be close and a bigger stake would recover the losses in a single move.
For several months the approach appeared to work, because most losing runs were short and each recovery trade wiped out the accumulated deficit. Then came a run of eight losing trades in a row, an outcome that was uncommon but entirely consistent with his roughly even win rate, and the doubling rule turned a manageable string of small losses into a position far larger than his risk limit allowed.
The illustrative firm responded not by punishing the trader but by hard-coding position limits into the trading system so that stake size could not increase after a loss. The chief risk officer's summary was blunt: the market does not owe anyone a winner, and any rule that assumes otherwise eventually meets a long streak.
Watch out
Common mistakes.
- Believing that independent outcomes must even out over the short term. They even out only over very large samples, and they do it by dilution rather than by correction.
- Increasing stake size or forecast confidence after a losing run. Doubling down on a supposedly overdue outcome is the fastest known route from a small loss to a serious one.
- Dismissing every long streak as pure chance. Sometimes a run really is evidence that your assumed probability is wrong, so the honest move is to test the assumption rather than to explain the streak away.
Questions
People also ask.
Does this only matter for gambling and trading?
No, it distorts sales forecasts, hiring decisions, quality control and insurance pricing wherever people read meaning into short sequences.
How can I tell if events are truly independent?
Ask whether the outcome of one attempt physically changes the conditions of the next, as drawing a card without replacement does and as tossing a coin does not.
What is the practical defence against it?
Write down the base rate before the sequence begins, then judge results against that figure rather than against your memory of the last few outcomes.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%