Tom Tests the Hot Hand

TrueHoop reader Tom has been very interested in this hot hand talk.

A big hang-up for Tom, and a lot of people has been the word that the groundbreaking recent study fixated on how likely a make was to be followed by a miss. As if two shots captured the ebbs and flows of shooting touch!

I could talk all day about how this is not in fact measuring two shots -- it's measuring every shot, and whether or not the makes and misses tend to come in groups -- but don't take my word for it.

TrueHoop reader Tom had a point similar to many people's: You need to define the hot hand in some way, perhaps as six of eight or whatever, and then see if getting hot in that way makes people better shooters for a while. (This would show up in John Huizinga and Sandy Weil's study, but no matter. Also, these kinds of studies have been done, didn't find the hot hand, and are often discredited for their rigid definition of a hot hand, which is a good reason not to do it that way.)

Anyway, Tom did his own micro-study, and here's what he found:

I chose Ben Gordon because he is a "streaky" shooter, he puts up a large number of shots with great consistency and he is largely a jump shooter. I looked at the start of every game he has played this season and defined a 'hot' start as a game where he started at least 4 out of 6 from the field and a 'cold' start as when he started 2 from 6 or worse.

Not necessarily perfect but not a bad system.

Over the season, Gordon had 27 games starting at least 4 of 6 (interestingly, none where he started 6 of 6) and 33 where he started 2 of 6 or worse (4 games of 0 of 6).

If the hot hand existed then it would be expected that on games where Gordon started hot, he would continue to shoot as well or at least at his long term field goal percentage. If cold, he would shoot as badly or no better than his long term field goal percentage.

In fact, the opposite happened. In the games Gordon started hot, he shot 43.3% for the remainder of his shots. In games he started cold he shot 48.9%.

His field foal percentage across the entirety of these games was 45.4%, as expected this is very similar to his season percentage of 45.5%.

This seems to confirm the evidence from the study; not only does making one shot lessen the probability of making the next shot and missing the previous shot enhance the probability of making the next shot, shooting well for several shots lessens the probability of shooting well for the rest of the game and starting badly increases the likelihood of shooting better for the rest of the game.

The seemingly obvious conclusion is that players who start well believe they are "hot" and consequently settle for worse shots/take harder shots and players who start poorly believe they are "cold" and consequently are more careful to take better shots.

Tom later did similar research into Ray Allen and Mo Williams. "Found very similar results," he e-mails. "No evidence of a hot hand, Allen was slightly worse both when hot and cold -- maybe he just shoots better at the start of games and Mo shot almost exactly the same after starting hot or cold -- maybe evidence that he is better at maintaining his view of what is a good shot."