The Clustering Illusion

The clustering illusion is the tendency to misinterpret "clumps" or "streaks" that naturally arise in a random distribution as non-random, that is, as meaningful patterns.

The cause lies in people underestimating the variability of randomness. Truly random data clumps and streaks far more often than we think. But because our brains expect that "if it is random, it should be spread out evenly," when we see a natural clump we feel that "something is going on."

During World War II, London residents believed a rumor that German bombs were targeting only certain areas. That was because some districts seemed to be hit especially often, while others appeared untouched.

But when the British actuary R. D. Clarke analyzed in 1946 the bomb hits across various districts of South London using the Poisson distribution (a statistical model for random events), the distribution matched a random scattering well. There was no evidence that certain districts were deliberately targeted.

In other words, the "pattern" people saw was simply the clumping that naturally arises from randomness.

In basketball, people believe that a player who has made several shots in a row has "gotten hot (a hot hand)" and will make the next shot too.

In 1985, Gilovich, Vallone, and Tversky analyzed NBA shooting records and reported that streaks of makes and misses were statistically indistinguishable from the random streaks produced by coin flips. Their conclusion was that the "hot hand" is a cognitive illusion of seeing random streaks as special.

However, this topic is still debated today. In 2018, Miller and Sanjurjo argued that the original study's statistics were biased, and that once corrected, a hot hand may exist to some degree. The core lesson still holds: before concluding that a streak is a "trend," you should examine whether it could arise from randomness as well.

Looking at charts, you see trends that are not there, and a fund manager or stock that has been right a few times in a row feels "hot." But even in a near-random market, streaks of success and clumped surges can readily appear.

Whether a streak of good performance is evidence of skill, or a clump that naturally arose from randomness, cannot be judged by eye. Just as the London residents did, you need enough data and statistical verification instead of intuition by sight.

This site does not point you to "hot stocks." Instead it shows you the actual long-term results, together with the maximum drawdown and the loss period, to help you avoid being swayed by short streaks.