How Important Is Individual Defensive Rebounding?

[Jacob Frankel]

Big men like Chris Bosh and Marc Gasol get a lot of flak for their below average DRB%s. The perception is that a team's rebounding is anchored by their big men, and below average rebounding big men hurt their teams rebounding. I did some research to see if this was actually true.



I looked at every big man (categorized by B/R) between the 2000-01 and 2010-11 seasons (676 players in all). I regressed these players' DRB% against their teams ' on-court DRB%. As you can see there's no correlation.

Interesting little tidbit I thought I'd share.

[Crow]

Image didn't post or at least does not appear for me. But thanks for sharing the findings. I had heard that general finding previously.

[v-zero]

Obvously defensive rebounding is very valuable, but as you rightly point out it isn't very significant to the individual. This is generally accepted to be true due to their being strong diminishing returns, which would appear to translate to: for most players doing some defensive rebounding isn't very difficult.

[DSMok1]

Obvously defensive rebounding is very valuable, but as you rightly point out it isn't very significant to the individual. This is generally accepted to be true due to their being strong diminishing returns, which would appear to translate to: for most players doing some defensive rebounding isn't very difficult.

I think there are two factors at work here (which have been discussed many times):
1. Many defensive rebounds aren't contested by the offense/they can go to anyone on D with no skill required. This is effectively "noise" compared to rebounding skill.
2. Much of defensive rebounding depends on assignment--players get more or less defensive rebounds based on what their roles are on the defensive boards. This also is effectively "noise" compared to rebounding skill.

Here was an adjusted rebounds study (using RAPM methodology): http://web.archive.org/web/201103111550 ... rebounding

Here is J.E.'s thread about that study: http://godismyjudgeok.com/DStats/APBRme ... 52%20.html

[Jacob Frankel]

Thanks for the links.

Another thing I'm sure you've discussed before is the massive disparity between the DRB%/playing time correlation and ORB%/playing time correlation, which strikes me as somewhat odd. Would you happen to have any links on that subject?

[Mike G]

Most guys with low minutes per game are coming off the bench. So they're playing with and against more than normal bench guys.
Non-starters miss more shots, so there are more rebounds available per minute. Presumably bench players don't box out on the defensive boards quite as well; and/or they're just more inclined to hit the offensive glass.

[Neil Paine]

Just to hit on some of the earlier points made by Daniel and V-Zero, defensive rebounding is a case where there are huge diminishing returns, as well as players who "steal" DReb opportunities from teammates:

http://insider.espn.go.com/nba/story/_/ ... umbers-nba
http://blog.philbirnbaum.com/2011/01/do ... nding.html

Do a little poking around on Google, and you'll find a lot of epic battles waged between Phil Birnbaum/Guy Molyneaux and the Wages of Wins crowd over this topic. The WoW people ran their regression between wins and boxscore stats at the team level, and -- big surprise -- defensive rebounds show up as very important (because at the team level, they represent defensive stops). But as Guy has hammered home (and I do mean hammered, you really should read some of these threads), adding an elite defensive rebounder to a new team makes barely any impact on the new team's defensive rebounding %, suggesting massive diminishing returns for individual DReb numbers. Another of Guy's favorite points is that a big-time defensive rebounder -- say, Rodman -- almost ALWAYS happens to play with teammates who, collectively, have a below-average defensive rebounding rate. Either this is an incredible coincidence, or most of the big DReb guys are just grabbing those boards because that's their role on the team, and their teammates have deliberately left them with a disproportionate amount of the team's DReb chances.

[Jacob Frankel]

Most guys with low minutes per game are coming off the bench. So they're playing with and against more than normal bench guys.
Non-starters miss more shots, so there are more rebounds available per minute. Presumably bench players don't box out on the defensive boards quite as well; and/or they're just more inclined to hit the offensive glass.

I'm using defensive rebound percentage (not DRR), so the more missed shots thing doesn't really explain it. Might just bench guys' style of play though.

Thanks for the links Neil. I'm new in these parts so any and all links to older discussions are appreciated.

One other note. While DRB% may not matter that much on the actual rebounds front of things, it had a higher than expected R^2 (0.25) when I regressed it with DRAPM. This may be more correlation than causation though.

[Mike G]

Jacob, you said you were using the DRb% as seen at b-ref.com, right?
I believe they still assume a constant rate of rebounds for a team and their opponent, for any minute of a season.
Therefore, if your minutes tend to be coming off the bench late in blowouts, for example, your Reb% has as its numerator the rebounds you got, and the denominator is the average for (team + opponent), over the whole season, for the minutes you played.

If your Reb% is from play-by-play, then you are more correct that your role doesn't matter. And even then, it should be adjusted to the personnel on the floor.

[Jacob Frankel]

Ah, my bad then.

[Mike G]

Rodman had more impact on teams' OReb% than on DRb%. He was on 3 teams in a 4 year interval:

Code: Select all

tm/yr    ORb%    DRb%
Det93    .346    .678
Det94    .284    .661
        
SA93     .278    .701
SA94     .356    .705
        
SA95     .309    .723
SA96     .275    .700
        
Chi95    .329    .682
Chi96    .369    .711
    
.averages    
Worm+    .345    .704
Worm-    .292    .686
diff     .053    .018

The '95 Spurs were a tad below average in OReb, as Rodman missed 33 games. Terry Cummings on a bad leg tried to fill in and had the best ORb% he'd have for the rest of his career.

[AcrossTheCourt]

Eli has a great post on the subject:
Diminishing returns.

Also, the Wins Produced formula was actually changed to reflect the effect of diminishing returns in rebounds.

http://wagesofwins.com/2011/12/11/wins- ... -stronger/

Look like they lost that battle. Hopefully they'll finally understand the usage versus efficiency topic.

[Mike G]

When you gain some insight, you haven't lost anything.

[v-zero]

Eli has a great post on the subject:
Diminishing returns.

Also, the Wins Produced formula was actually changed to reflect the effect of diminishing returns in rebounds.

http://wagesofwins.com/2011/12/11/wins- ... -stronger/

Look like they lost that battle. Hopefully they'll finally understand the usage versus efficiency topic.

Nobody yet grasps this really.

[Guy]

Also, the Wins Produced formula was actually changed to reflect the effect of diminishing returns in rebounds. Look like they lost that battle. Hopefully they'll finally understand the usage versus efficiency topic.

Well, the formula was tweaked. It reduced the value of DREB, but not nearly enough, and didn't change the value of OREB at all (which also have diminishing returns, though significantly less). So after seven years, they fixed maybe 30% of their rebound problem. Given that pace of reckoning with reality, I wouldn't hold your breath on usage.

Berri is also pretty dug in on usage. See this post for example: http://wagesofwins.com/2013/04/01/how-d ... ter-story/, in which Berri notes that the Rockets have the same # of FGA with and without Harden, and therefore there is no such thing as shot creation. Seriously. After years of debating the issue, it appears that Berri doesn't even have the faintest idea what this discussion is about.