0

Evaluating D-League Shot Selection with XPPS

Apr 30, 2014; Houston, TX, USA; Houston Rockets guard Troy Daniels (30) reacts after making a basket during the second quarter against the Portland Trail Blazers in game five of the first round of the 2014 NBA Playoffs at Toyota Center. Mandatory Credit: Troy Taormina-USA TODAY Sports

Last Monday, I brought Expected Points Per Shot (XPPS) to Nylon Calculus for the first time. XPPS is a rough measure of the quality of a team’s or player’s shot selection, built on that team’s or player’s individual shot distribution and the expected value1 of shots from different locations. We now have data available for XPPS and XPPS Allowed (measuring the quality of shot selection a team allows their opponents) for NBA teams going back to the 2000-2001 season. Today we’re adding to that database with something exciting — XPPS and XPPS Allowed for D-League teams, going back to 2007-2008.

Since XPPS is based on the expected value of shots from different locations, the first thing needed to make these calculations for the D-League was to find those expected values. Although they were fairly similar to the NBA, the point per shot averages were slightly different in some key ways.

Screen Shot 2014-08-08 at 12.04.10 AM

The differences between the values are small—just fractions of a point per shot. But what they show is that, except trips to the free throw line, the shot attempts that are most valuable in the NBA, are even more so in the D-League. This leads to an interesting scenario where the average shot distribution in the D-League is actually more efficient than that of the NBA. The table below shows the average XPPS for each league over the past few seasons.

Screen Shot 2014-08-08 at 12.17.18 AM

The big swings in the D-League XPPS from season to season can be mostly attributed to the fact that there are far fewer teams in the D-League. But over the past seven seasons the expected value of the D-League’s shot selection has been consistently and dramatically higher than that of the NBA. This is coming, not just from the higher expected value of shots at the rim and corner three-pointers, it’s also from the distribution of shots. The table below shows the average distribution of true shot attempts (field goals and trips to the free throw line) for each league from last season.


Screen Shot 2014-08-13 at 1.42.05 PM

The first thought I had when looking at these numbers is that they were unfairly skewed by the science experiment going on with the Rio Grande Valley Vipers. Although the Vipers do have six of the eight highest Team XPPS marks in the D-League data (all of which were higher than the Houston Rockets’ XPPS last season—the highest NBA XPPS of all time), when I removed their numbers the changes to the D-League average distribution was negligible.

The way these numbers play out is an interesting manifestation of the differences between the NBA and the D-League. The skill level is the NBA is obviously higher and implemented more cohesively given how much more practice time they have and how much more roster and front office consistency there is from year-to-year. This seems to impair defensive consistency in the D-League as teams get more value from easy shots and appear able to get those easier shots more regularly.

Our XPPS dataset is still expanding here at Nylon Calculus and over the next few weeks we hope to have several seasons of individual player data available for both the NBA and D-League. In the meantime, have fun playing with the numbers and let us know what you find!

  1. League average going back to 2000-2001

Ian Levy

Ian Levy (@HickoryHigh) writes about basketball from the wilds of Southern Vermont. In addition to his work for Hardwood Paroxysm, he is the man behind the curtain at Hickory-High and a contributor to Indy Cornrows, The Two Man Game and HoopChalk.