Lebron James sure had the “hot hand” last night. He scored 61 points on 66% shooting and 80% on three pointers. After the game, James said:
“It felt like I had a golf ball, throwing it into the ocean.”
Was he really “feeling” that way? Did he really have the “hot hand”? Or was this just a case of our biases messing with our perception of reality?
Over the last 40 years Daniel Kahneman and Amos Tversky have put a good deal of doubt in this idea of the “hot hand”. They referred to the idea as the “gambler’s fallacy”. This is the idea that you can predict the future by understanding the recent past. Of course, this is true for a game of pure chance like roulette where the game is always the same and the outcomes aren’t influenced by any outside variables. But what about for a game of skill? Researchers had performed studies showing that a game of basketball exhibits similar traits where the “hot hand” appears to be a fallacy. But a new study presented at this week’s MIT Sloan Sports Analytics Conference is putting that 40 year old conclusion in doubt. The author’s find that a game of skill like basketball does display elements of the “hot hand” effect.
This reminds me a bit of the Efficient Market Hypothesis debates between traders and academics. Any trader who has actually been involved in markets is likely to reject the EMH based almost entirely on experience whereas an academic can simply reject this by looking at data. Likewise, anyone who has played basketball knows that some days you just don’t “feel” it. It could be a lot of different factors, but basketball is not a game of pure chance. So it’s not surprising, to athletes, that the “hot hand” could exist, just a trader is likely to reject the EMH.
But the important question is not really whether it exists, but whether you should change your behavior because of it? In a game where there is a degree of skill involved, should we approach the game differently at different times based on the perception that someone has the “hot hand”? In my view, this would be misguided. For instance, Lebron James should not alter his playing style just because he is “feeling” it. If he does, he could be inclined to take risks (for instance, more difficult shots) that he normally does not which could actually harm his “hot hand”. Likewise, defenders should play Lebron just as though he’s his standard 25/7/7 self. After all, we know that while he might have the hot hand, the odds do not favor playing him as though he will sustain his “hot hand”. And over the course of an entire NBA season (or even a few games), it would be detrimental to approach him as though he’s able to sustain such a high level of play.
In sum, while the “hot hand” likely exists, I am not sure it changes much in terms of the way we should approach games of skill. In fact, approaching the game differently based on the “hot hand” effect could negatively influence our future outcomes. When we have long data sets that can improve our understanding of potential long-term outcomes, recency bias and the reach for short-term gain can negatively influence our potential for long-term success.
* (Thanks to Bradley Parkes for bringing this article to my attention)