NBA Analytics and the Rise of the Robots

“Analytics” has been the buzzword in basketball for a very long time. And for a very long time, I have expressed some criticism of the “analytics” people have offered.  If you bother to peruse the 1,000 plus posts on this blog, you will likely see some of this criticism.  Or if you don’t have the time, Chris Benjamin* has written an article for Men’s Journal — The 4 Fallacies of NBA Analyticsthat details much of my critique.

The list includes decision-makers focusing too much on visual observation and scoring, collecting data without knowing how to connect that data to outcomes, and creating measures that are entirely ad hoc and not capable of explaining wins.  Benjamin’s piece captures in a few paragraphs much of what I have said the past nine years.

Benjamin’s article also gives me a chance to re-state the role I think analytics should play in decision-making.  The following was said in The Wages of Wins:

These stories demonstrate that one cannot end the analysis when one has measured the value of player performance. Knowing the value of each player is only the starting point of analysis. The next step is determining why the player is productive or unproductive. In our view, this is where coaching should begin. We think we can offer a reasonable measure of a player’s productivity. Although we have offered some insights into why players are productive, ultimately this question can only be answered by additional scrutiny into the age and injury status of the player, the construction of a team, and the roles the player plays on the floor.

In sum, Wins Produced is a reasonable measure of how productive a player is on the court.  The model seems to do a very good job of explaining how each player impacts outcomes.  And if performance never changes, Wins Produced is all you would ever need. But because player performance is impacted by age, injury, productivity of teammates, etc… there is a need for more analysis.

Part of that analysis has been offered (both in The Wages of Wins and Stumbling on Wins).  In these books the role of age, coaching, teammate productivity (with respect to assists and defensive rebounds) were noted.  Predicting performance in the future requires that you at least take this into account.  And because Wins Produced includes the impact of teammate assists and teammate defensive rebounds, before you would even try to predict anything with Wins Produced you would have to essentially unpack Wins Produced and re-construct the measure with what those impacts would be in the future (and make also consider the impact of player position).  In other words, because performance is not constant and some of what causes this to change is incorporated in Wins Produced, you can’t just take last year’s Wins Produced and try to predict next year’s Wins Produced.  Well, you can.  But you are not really doing much that is useful.

And I would add, you probably shouldn’t – as The Wages of Wins clearly states – simply rely on numbers in predicting future performance and making decisions.  Again, there is a role for coaches and other decision-makers.  But their role should begin after a player’s past value has been empirically measured.

Despite this statement, Professor Phillip Maymin – an Assistant Professor of Finance and Risk Engineering at the NYU School of Engineering – recently asked a bold question:  What if the Knicks and Nets were run by Robots?   As Professor Maymin argued in a recent paper, a computer could draft players better than NBA general managers (Stacey Brook, Aju Fenn, and I also made a similar argument in a published paper).  So why not just take the general managers out of the equation?

As I recently noted in following two articles at the Huffington Post

“Bad Decision Making is a Pattern With the New York Knicks.”

“Isiah Thomas is Simply Not Good Enough at His Job to Get a Pass on His Past Behavior.”

that the Knicks are especially bad at making decisions.  And naming Isiah Thomas to run the New York Liberty of the WNBA is just the latest example.   In addition, I have also noted in the past the the Brooklyn Nets are not exactly an example of great decision-making.

But although both the Knicks and Nets need to make better decisions to find success, I think a person who actually understood analytics would do better than a computer.  But in the NBA – as argued in Benajamin’s article – I am not sure everyone really understands all the numbers being thrown around.  And until that happens, maybe teams should turn to Professor Maymin’s computer.

– DJ

* – Chris Benjamin is an award-winning author and freelancer who writes for magazines, newspapers and online publications across Canada and the US. His new book, Indian School Road: Legacies of the Shubenacadie Residential School, won the Dave Greber Freelance Award. He won a silver Atlantic Journalism Award for his piece, “Betting the Fish Farm,” in the Summer 2014 issue of Saltscapes Magazine.