Freelance Friday is a project that lets us share our platform with the multitude of talented writers and basketball analysts who aren’t part of our regular staff of contributors. As part of that series we’re proud to present this guest post from Mika Honkasalo, a 21-year old NBA enthusiast from Helsinki, Finland who is studying computer science and mathematics. Mike enjoys big men who can pass and players who can shoot off of screens, find him on Twitter @mhonkasaloNBA.
If you haven’t seen it by now (and are taking time to read this column), you should probably go over and watch the latest episode of Open Court on NBA TV, on which the use of analytics in the NBA got mauled.
Or to be more precise absolutely destroyed. Passed off as a joke. As an invention of people who don’t understand or watch basketball. Stuck, in their video game world.
"“If we want to be honest, it’s about people validating unwarranted knowledge” — Chris Webber"
And while we can break down every statement made and examine all the misrepresentations, exaggerations and distortions line-by-line. The real problem seems to be the very common misconceptions about analytics that all of us, who enjoy this sort of thing, run into constantly.
The problem really starts with the use of the word “analytics,” typically defined as the method of logical analysis, or more strictly the communication and discovery of meaningful patterns in data sets.
“Analytics” really is a suitcase term like “sports”. To illustrate this point, let’s take the contrast between ice hockey and bowling. We use the word “sports” to describe both of these activities, yet they have next to nothing in common. One is fast paced, physical, full of contact. The other is not. The two have close to nothing in common, save the fact that they are generally played by people who all breath air.
The term analytics is very much the same. There are counting stats, big data sets, predictive models and a whole slew of other ways to organize and analyze data depending on what you want to learn. While these tend to involve math and computing, the mindset is very similar to any other type of analysis one might exercise themselves in, whether it be watching tape on Tony Parker pick-and-rolls or interviewing potential draft candidates.
The goal should be to acquire more information, understand and analyze it before applying it to real life decisions. We should all be open-minded and eager to learn new things. And I would assume that’s the kind of people you want running NBA teams.
When people complain about and dismiss analytics, their retorts tend to be in a few categories.
The first is that analytics is only an exercise in reductionism and that it doesn’t add any value. Ernie Johnson was the one to paint this picture when he brought up the formula to calculate Player Impact Estimate. First of all PIE is not “advanced” at all considering you only need to know second grade math. And I’m sure we’d all agree if there ever was a holy grail of a single number stat, PIE would not be it.
Moreover, this suggests that the point of analytics is to simplify the game to a few numbers, and that we simplify because we don’t understand.
Of course professional athletes know more about their field than anyone else. But it’s not intellectually honest to think that everything in a complex systems such as NBA games can be understood just by watching the game, only via the “eye-test.”
Take for example the thought experiment posed in the “Pointwise: Predicting Points and Valuing Decisions in Real Time” presentation at last year’s Sloan Conference Kawhi Leonard hit a game winning shot from the corner against the Cavs and the question was raised: “How many of you can tell me all five players the Spurs had on the court right now?”
Expanding this idea, unless you can tell me at every single point of the game where each player is and who they are, you are by definition missing some data. Add to the mix this Kirk Goldsberry graph that depicts shot values in expected points per shot which shows us that every shot between three feet and the three-point line is worth, on average, between 0.75 and 0.81 points. That is a fact I bet is counter-intuitive to most NBA people who are watching the game. It may look terrifying when Carlos Boozer backs down a smaller guy, but unless he can get all the way to the basket the resulting shot probably isn’t a really efficient one.
This shows our intuitive picture of the game can be and is often both incomplete and incorrect. That’s the whole point of adding science to the mix! If everything in the world came to us intuitively we wouldn’t need math or science.
Analytics tends to be misrepresented because it is a field that is often both understood and communicated poorly. “Kevin Love is third in PER, he’s awesome” and “you can’t put a player’s value into one stat” are lines you might hear on opposing sides of an NBA argument. Legitimate criticisms of PER might be that it doesn’t measure defense or value efficiency correctly, but the distinction is that PER doesn’t pretend to do those things.
It’s said that you can’t quantify heart, effort and team chemistry. The keyword here being quantify. Is anyone actually trying to? Yet it doesn’t mean that team chemistry doesn’t show up in the data, and that you can’t look for the value of things like continuity. I’m pretty sure effort and heart actually do show up on the “stat sheet,” they’re wrapped up in many other aspects of player performance.
Another common criticism is that numbers take some enjoyment out of the game. Richard Feynman made a great point about this, which is that understanding how a rainbow works doesn’t make it any less beautiful.
Why would it make you appreciate Dirk any less if you know that he shoots say, 48 percent when he fades by turning towards his right shoulder and 41 percent on his left?
This probably ties into the fact that people think math is boring and in a sense, scary. Producing results set in stone, bereft of interpretation and showing a lack of imagination. They say numbers lack context. Which is why I’m not a fan of sayings such as “numbers don’t tell the whole story” or “stats are just a part of analysis”. Not because they’re not true, they certainly are. It’s just that it makes it look like analyzing data is an action that produces results set in stone — unimaginative and inflexible.
It’s actually as fluid of an analysis as the eye-test. To me, it’s what also separates a smart analytics person.
Just like anyone can look at the result of the game or how far a team got in the playoffs and make decisions to fire the coach or trade players, anyone can look at a number and say that player X is worth Y, but it takes a really smart analytics person to have the imagination to project what might be.
Take the example of Boris Diaw. Diaw had negative Win Shares with the Charlotte Bobcats during the 2011-12 season, almost an impressive feat. Two years later he’s starting in the finals on an iconic Spurs team en route to their fifth championship during the Tim Duncan era. Instead of looking at his Win Shares and concluding that Diaw was done because players of his age rarely bounce back from such a dip in production — the Spurs saw a player who could thrive in their system.
Does any of this really matter? Maybe. The role of analytics in sports is going continue to grow regardless. On the other hand it’s been stated that people doing analytics for NBA teams have a hard time communicating with “basketball” people. Obviously I don’t know how true this is since I don’t work for an NBA team. What seems clear though is that if we can’t create the type of open dialogue where analytics are discussed in a way that makes it more approachable, where both parties are open to each others views, we’ll always be stuck on the margins as the nerds who “just don’t get it”.
When people who follow the NBA think about analytics they think Box Score, PER, Win Shares and Real Plus Minus. Focus should be less on that, and more about the critical thinking applied to solving interesting problems.
*A note about myself. I used to play ice hockey at the national league level for juniors in Finland. While I wasn’t good enough to make it pro, I played with/against a lot of Finns who did make it. Hockey is sort of an “old-school” culture, and I can understand the criticism that analysts don’t get how hard you have to work and the mentality you need to put yourself in to succeed. I do however believe that in that sphere, I could’ve learned a lot from analytics.