Understanding the hot hand, the myth of the hot hand, and the myth of the myth | Hacker News Reader

Understanding the hot hand, the myth of the hot hand, and the myth of the myth

andrewgelman.com

33 pointsarch_stanton8 comments

Here's some context:

http://en.m.wikipedia.org/wiki/Hot-hand_fallacy

Except the argument in sports is exactly this:

   > Athletes are not machines, and anything that can 
   > affect their expectations (for example, success in 
   > previous tries) should affect their performance—one 
   > way or another. To put it another way, there is little 
   > debate that a “cold hand” can exist: It is no surprise 
   > that a player will be less successful if he or she is 
   > sick, or injured, or playing against excellent 
   > defense. Occasional periods of poor performance will 
   > manifest themselves as a small positive time 
   > correlation when data are aggregated.

Basically the question of what factors could be present that would increase the success likely hood of a basketball scoring attempt, and then whether or not those factors can persist over time or if they are always just noise. And the myth part being that someone who believes they will be successful is more likely to succeed than someone who has no expectation of success.

The science then is can the expectation help, and if so can it bias success in a statistically significant way.

> (3) not every streak is a hot streak, so the real infrequent, but persistent hot hands get diluted;

no true scotsman would have a not-hot streak.

On a purely mathematical level, though, you can see how this could be true.

Like, take basketball out of the equation. Let's say I code up a button. When you press it, it shows a screen that says "You scored!" or "You failed to score. :(" Now, I code the button to generally say "you scored" 1/2 of the time, and "you failed to score :(" 1/2 of the time. But I also code it to say that on 1% of all scores, it enters a "hot streak" mode where for the next ten button presses, it's 80% likely to score.

We would agree, yes, that this button has hot streaks?

But now suppose that you look across a large dataset for periods of ten button pushes in which more than 7 of them are scores. Those are streaks of scores. Some of those are going to be because the button is in hot streak mode. But some of them are just going to be because the normal 50% chance clustered.

Now, is my button anything like a good model for how basketball works? I have no idea. But the sentence "not all streaks are hot streaks" is not self-evidently crazy.

Or to pick at the semantics a bit: Not all streaks indicate causality. Some are random artifacts.

Easy enough to prove by flipping a coin. Streaks naturally happen and should be expected from randomness.

I spent a long part of this article thinking it was about poker.

as with most real world situations, the problem (in my estimation) is that statistics and the experimental model are not yet sophisticated enough to sort out the effect, so it's hard to distinguish from random. the statistical model would likely need many more variables, be non-linear, and measure things (like emotional positivity) more dynamically and/or accurately than possible today.

for example, my shooting shoulder is chronically injured. if i warm up properly, i seem to shoot better. is this a psychological effect? a physiological effect? something else? in game situations, it's hard to isolate the effect from all the confounding variables. even if you could, you'd need to measure all the confounding variables to negate them in the analysis.

i suspect a non-logic based "math" will have to be developed (quantum computing?) before we can understand these kinds of situations.

It appears academics argue much like folks on the Internet argue, just with different vocabulary and maybe a touch nicer.

Reply on news.ycombinator.com