Hot-hand fallacy
Also called hot hand
Believing a streak of successes makes another one more likely. In casino games it is an error. A 1985 basketball study called it a fallacy, but Miller and Sanjurjo (2018) showed its test was biased and found real streak shooting.

- Level
- Intermediate
- Reading
- 2 min
- Sources
- 2Research
- Updated
On this page· 5 sections
A basketball shooter hits three in a row. Fans, coaches and players say the shooter is “hot” and more likely to hit the next one. For decades, researchers called that belief a mistake[1].
In 2018, a statistical analysis of the original test overturned the verdict. The story is useful to a trader for what it says about streaks and small samples.
How it works
The fallacy comes from a 1985 study of basketball shooting by Gilovich, Vallone and Tversky. Despite the near-universal belief in momentum among fans and experts, its statistical tests concluded that momentum in shooting did not exist[1]. The hot hand came to be treated as a “myth”[1].
In 2018, Joshua Miller and Adam Sanjurjo showed that the test behind the verdict is biased. It looks at what happens right after a streak of successes, and on a finite sequence that selection method understates the true chance of another success[1]. Their re-analysis of the 1985 data, with the bias corrected, found statistical evidence of streak shooting, with large effect sizes[1].
The belief is also studied in gambling. A field study of casino patrons filmed on video found small but measurable biases, consistent with lab results, in both the gambler’s fallacy and the hot hand[2]. Miller and Sanjurjo’s paper also notes that the fallacy has been used to explain puzzles in financial markets, sports wagering and lotteries[1].
Example
Say you flip a fair coin three times and write down every flip that comes right after a head. Then you compute the share of heads among the recorded flips. There are eight possible sequences and six of them record at least one flip.
| Sequence | Share of heads recorded |
|---|---|
| THT, HTT, HTH | 0 |
| HHT | 1/2 |
| THH, HHH | 1 |
The average is (0 + 0 + 0 + 0.5 + 1 + 1) ÷ 6 = 2.5 ÷ 6 = 41.7%. On a fair coin, the share of heads after a head looks lower than 50% in short sequences[1].
Watch out
The label depends on the process. Fair coin flips are independent by design, so a streak predicts nothing there. For skilled performance such as shooting, the 2018 analysis found evidence that streaks exist[1].
A bias in the test is not proof of a hot hand in every setting. Neither paper studies traders or trading results.
The belief can fail in two directions. People who expect a streak to continue and people who expect it to reverse both depart from the laws of chance[2].
Sources
- ResearcharXivSurprised by the Hot Hand Fallacy? A Truth in the Law of Small Numbers (Miller & Sanjurjo, Econometrica 86(6), 2018)Checked November 9, 2026
- ResearchRePEc IDEAS (abstract record)The Gambler's Fallacy and the Hot Hand: Empirical Data from Casinos (Croson & Sundali, Journal of Risk and Uncertainty 30(3), 2005)Checked November 9, 2026
