With sincerest apologies readers, we were duped on this one. Evan Zamir was making a "what-if" interview with Kahneman. Of course, this was very hard to derive as the only "disclaimer" was:
The following interview with Nobel Prize winner Daniel Kahneman may have taken place.
This was immediately followed by:
I recently was given an opportunity to sit down with Dr. Daniel Kahneman who won the Nobel Prize in Economic Science back in 2002.
As such, all statements this piece made originally in regards to Kahneman's "missteps" in basketball knowledge have been removed. We find impersonating a major name in an interview to be a disingenous act on the part of Evan Zamir and Golden State of Mind, and apologize for any perpetuation in that misstep. The article has been updated to remove any incorrect references to Kahneman.
We love Daniel Kahneman around here. Thinking Fast and Slow is a must-read. The idea that how we process information impacts our decisions is popular amongst the geeks. So when I heard that Dr. Kahneman had talked about basketball, my ears perked up. Of course, the article was actually Evan Zamir imagining what Kahneman would say. This included things like Kahneman liking Zamir's site and following him on Twitter. Additionally, it showed that Zamir misunderstood Kahneman's work. As such, we were less than happy with the "Fake Kahneman" provided.
The Law of Small Numbers
One of the first things that surprised me was when Zamir assumed that Kahneman would use players' on/off court stats. On/off court data and lineup analysis are super popular, but the sheer number of permutations and variables for small segments of gametime make it ridiculous to use them to draw conclusions. I often see people say "with David Lee on the court, x , but when he's off the court, y ". The issue is that other variables have shifted as well. Who replaced David Lee? Who are the Warriors playing? Did the Warriors' opponent make a substitution at the same time? There are tons of these variables and the sample sizes used in on/off court data are just too small to draw many meaningful conclusions. So would Kahneman really like these?
Of course not! It turns out that, along with Amis Tversky, Kahneman has written about the Law of Small Numbers. In essence, people are prone to use samples that are far too small to be conclusive, and furthermore, they are prone to put far too much faith in the outcomes of these small samples. It's nice to know that the problem with plus-minus stats were being explained over thirty years ago! Zamir apparently didn't read that chapter in "Thinking Fast and Slow".
Evan intuited that Kahneman may have thought the following about Klay Thompson
Well, let's take Klay Thompson. This kid is a prolific and productive shooter. He's Jerry West's "golden boy".
I was a bit confused by this comment. This season Klay Thompson's True Shooting Percentage (TS%) is around 56.4%; understand that this is the best mark of his career by a large margin. Career-wise he has a TS% of 54.5. If we use Basketball-Reference's Play Index, we can find over 300 players who have taken over 15 shots a game and hit this same percentage. Even when we limit this search to just this season, we still get 11 players!
When it comes to shooting, one stat I like to look at is Expected Points Per Shot (developed by Ian Levy). This examines the quality of the shots players take. Thompson has shown that he hits above the expected rate for his shots, but he takes poor shots. If we examine his shot charts from this season and the last, we can see that, while Klay shoots a lot of threes (which is good), he's also in love with midrange shots (which is bad). While I can accept the claim, I'm not a fan of calling someone who takes so many ill-advised shots "productive".
So I don't think that Evan or Kahneman should think Klay is prolific. I'm certainly confused as to why Evan would believe that Kahneman would think such a thing.
In his failures to impersonate a great mind, Zamir did have one good point. Klay Thompson is often called inconsistent. The reason for this is pretty simple: Klay takes a lot of threes. I have no qualms about calling Klay Thompson "prolific" when it comes to threes; if his season ended today, he would rank 11th all-time in 3PT% among players who take more than six three pointers a game. [Fun side note: he's just behind Damian Lillard and Wesley Matthews, who are both on this year's Blazers!]
The issue is that – even at 40% – we can expect Thompson to be all over the place in individual games. Case in point: while he averages 19 points per game, he's scored as high as 38 points in a game this season, and as low as 5. Evan's fictional Kahneman did a fun experiment though. If Klay's shooting followed a poisson distribution, what would it look like? How does reality compare? Well, the reality is that Klay's shooting looks more like a normal distribution. Thus, he's not "inconsistent", he's just following the normal distribution you'd expect for someone with his shooting percentage.
And I loved this point. I hear a lot about "consistency" in players. The thing is that players are all over the place – even greats like LeBron. Players like Klay, who shoot a lot of threes, will appear more inconsistent, but that's not true. And this is something that I wish fans and coaches would pay more attention to: a good player can have a bad half, and a bad player can have a good half. It doesn't mean you should bench them, play them, trade them, etc., or any other irrational moves that we see front offices make!
Way back when, David Berri's research ended up in the middle of an intellectual scuffle with Steven Pinker. Pinker has a background in cognitive psychology and philosophy. On the other hand, Berri is one of the most published sports economists on the planet! Even still, Pinker took issue with Dave's work and argued against him!
Why bring this up? The simple fact is there's a big difference between being a fan of something and being an expert in it. Evan is clearly a fan of both the NBA and Kahneman. However, this does not mean he is an expert in either Kahneman's work or basketball! [To his credit, I have been impressed with some of Evan's software development skills, use of R, and his biology background seems good.] However, I've often found that people with strong backgrounds in one area (Evan has a PhD in Biology) like to assume that means they're experts in other realms. As the above article shows, this is often not the case.