How To Measure Rebounding

One of the greatest contributions of the “analytics revolution” in the NBA is that we no longer need to hear about how great a player is due to their points per game average. The broadcaster on tv, the journalist in your paper or your over confident loud mouth friend no longer advocate for the Michael Beasley, Andrea Bargnani or Rudy Gays of the world just because they can average 20+ PPG. While the basketball community at large has recognized the concept of efficiency and shunned points per game, rebounds per game is still ubiquitous.

While Reggie Evans 11 rebound per game season never made him an all-star candidate, a player’s rebound stats still have a huge influence on their reputation and value. What’s ironic about people dropping PPG while keeping RPG is that not only does scoring actually take talent in the NBA, but a player needs to convince his coach and teammates that them chucking is in their team’s strategic interest. Meanwhile, a players rebounding numbers is largely a function of how tall a player is and how many minutes the play.

One of the reasons why RPG is still lingering around is because there is no ideal rebounding stat that can suddenly make people realize how stupid RPG is (similar to how merely looking at missed shots can ruin someone’s view of PPG). For those not immersed in the field, NBA offices and the basketball analytics community have largely moved away from rebounding statistics and box-score based stats as a whole. This is because player evaluation is now primarily viewed through the lens of plus-minus stats that ignore individual box score metrics such as rebounding or blocks.

While I think plus-minus stats are superior to box score stats a whole, I still think box score stats have value. Rebounding is hugely important for an NBA team to succeed and if a basketball team wants to get better at rebounding, the easiest way to do that is to add better rebounders to their team.

When basketball analytics-savvy individuals want to look at rebounding, they normally look at rebound percentage. This metric measures the percentage of rebounds a player recovered out of all of the possible rebounds that took place while he is in the game. Another stat used is contested rebounding percentage, and this measures the percentage of rebounds a player collects that were contested by a player of the opposing team.

Some issues with these stats are the following:

  • –       does not take into account amount of possible rebounds (based either on minutes played, or missed shots in that time period)
  • –       does not take into account the rebounding abilities of teammates and other players on the floor
  • –       does not take into account what a teams strategy is (ie some teams do not attempt to get offensive rebounds)
  • –       does not account for boxing out to allow other teammates to grab rebounds
  • –       does not take into account the ability for a player to contest a rebound

Based on these flaws, a new and better rebounding stat would need to take into account the following factors:

  • –       would need to be based on per possession results rather than per game totals
  • –       would need to take into account not only a player’s teammates, but how good the rebounders are on opposing teams
  • –       would need to factor in how many additional rebounds the team gets when that player is on (adjusted for the quality of teammates and opponents)

Together, you have in my mind the perfect rebounding stat. By focussing on how many rebounds a player contributes to their team per possession with the quality of teammates/opponents regularized, all of the flaws with other rebounding stats are mitigated.

In order to make this calculation, we need to determine how many rebounds a player’s team recovered per possession while that player is on the floor in comparison to how many the team recovered while the player is not playing. Then we need to figure out how good at rebounding the nine other players on the floor were compared to average NBA rebounder while that player was on the floor. This can be calculated by running the same calculation for all players in the NBA for every possession played throughout the last several seasons.

Fortunately for me, I was able to convince Jeremias Engelmann, formerly of the Phoenix Suns and creator of RAPM (the most popular NBA analytic stat) to run these numbers for the 2015 NBA season. You can find the data here under the column ind. def. rebounds per 100 opp.

Reggie Evans ranks in first place and adds 30.14 defensive rebounds to his team per 100 opportunities while DJ Augustin finishes last of players in a regular NBA rotation with only 7.83 contributed defensive rebounds to his team per 100 possessions.

Because I’m a Raptor homer, I’ll list some of our players. Jonas ranks as the 33rd best rebounder in the league with 20.76 defensive rebounds per 100 posessions contributed. Chuck Hayes is next at 16.74, Stiemsma at 16.69, Amir at 16.46, Hansborough at 15.74, patterson at 15.29, james johnson at 14.08 , lowry at 13.55, ross at 11.57, fields at 13.70, demar at 11.04,