Player Tracking Plus-Minus: 2015-16

I am belatedly getting around to releasing the final Player Tracking Plus-Minus (PT-PM) numbers for 2015-2016. PT-PM is my version of a statistical plus-minus model, that I have developed over the last few years. It is based on some box score metrics (basic and advanced) that you’ll find in several other statistical plus-minus models but it also includes some metrics from the NBA’s SportVU player tracking statistics.

I wanted to have a place to reference the numbers for future posts and put them out there for anyone who is interested in seeing the full data set. First, a quick visualization, that pits the Defensive PT-PM against the Offensive PT-PM, with the color representing the overall score and the size of each player’s marker determined by the number of minutes he played over the season. The visualization is filtered for players with at least 300 minutes. Plus I think it looks kinda cool.

Players higher on the graph were better defenders. Players further to the right on the graph were better offensive players. The darker shade of a blue for each player’s marker represents a greater estimated net impact.

Finally, here is a table with the raw Offensive and Defensive PT-PM numbers as well as the total score adjusted for playing time (lower minutes players are regressed more to account for the small sample size). Table can be filtered by team or minutes played.

As you would expect, Stephen Curry was estimated to be the best player in the league last season by a fairly wide margin. His offensive impact dwarfed that of anyone else in the league. Of players who played at least 1,000 minutes, Kawhi Leonard was estimated to be the best defender, with an impact roughly one point per 100 possessions greater than Draymond Green’s. The top seven defensive players by PT-PM all played for either the Warriors or the Spurs.

Stay tuned for more work with PT-PM this coming season.