APBRmetrics

[EdKupfer]

All right, just so I have some content, here's a quick regression. People who know me well know that I feel players rebounding stats are totally overvalued -- especially defensive rebounds. Here's why:



The X axis shows player DR% numbers from 03-04 (1000+ minutes). The Y axis is the team DR% while that player was on the floor. The numbers come from 82games.com. You can see that the slope of the regression line is very flat -- for every 1% increase in player DR%, tha team only increases its DR% by 0.08%

Take a look now at OR%:



For every 1% increase in player OR%, tha team only increases its OR% by 0.22% -- almost 3 times the effect. (I've tried to show both graphs along a similar scale so you can compare the effect visually.)

Whatever you think rebounding in general is worth (and given the small slope on both graphs, I don't think individual rebounding is worth much), it seems pretty clear to me that lumping offensive and defensive rebounds together is a mistake.
_________________
--
ed

[edo]

Ed: thanks for posting. I am kind of new to this list, and so I have not had a chance to hear much about your opinion of the value of rebounds.

Are you saying that an individual's rebounding (especially defensive) ability is generally less valuable to the team since a team tends to rebound (or not) independent of any single player? I think this is what you're saying.

Ed O.

[EdKupfer]

[admin]

This should have what you're looking for, EdK.

How well an individual player's rebounding translates at the team level is a topic I don't think has gotten enough discussion, so I'm glad to see these charts. They would seem pretty strong ammo for the offensive rebounds more valuable than defensive rebounds school of thought (and, in fact, the three to one ratio you find is what I use in my linear-weights stuff).

At the same time, unless I'm misreading what you're saying, I think there's a very good reason those graphs are as flat as they are -- you're lumping together players of different positions.

Let's say we have a point guard who grabs 9% of defensive rebounds -- what I have as average for point guards at whatever point I did my study during the 2003-04 season. On average, we can expect his team would have a defensive rebounding rate right at league average, since that's where he's at.

Now let's compare him to a power forward who grabs 19.3% of defensive rebounds. That's a full 10% more and more than twice as many as the point guard ... but it's still league average for his position and still implies an expectation of a league-average defensive rebounding team.

On your graph, then, you see a player at 9% and a player at 19% whose team rebounding is the same, which appears to imply that the 10% difference is meaningless. In reality, however, it's not ... if you for some reason played the point guard at power forward, and he somehow managed not to have any difficulty whatsoever defensively, the rebounding would still obviously fall off because you would be replacing the 19% rebounding player with someone around 9%. Does that make sense?

I'd like to see the same chart separated out by position. ...

[EdKupfer]

[Guest]

Interesting data that supports what would seem to be basic logic. As you put it, a missed potential defensive rebound will still likely be grabbed by a teammate, while the same doesn't hold true for potential offensive rebounds. I'm just curious, but what set of data did you use? I suspect that the relative importance and effect of an offensive rebound has increased over the years with the decrease of possessions per game, and the more prevelant strategy to send defense back to prevent transition points rather than attack the offensive boards.

[Yyzlin]

Sorry. The above post was by me.

[EdKupfer]

[admin]

[WizardsKev]

I'd been resisting the "offensive rebounds are more valuable than defensive rebounds" thing for a little while -- primarily because the best offensive rebounding teams were often bad teams. But, often the "best" offensive rebounding teams get a lot of offensive rebounds because they're among the worst shooting teams, which is a key reason why they're also bad teams. Geez, you'd think all this stuff was connected or something.

[EdKupfer]

[admin]

Offensive rebounds show as slightly more valuable at the team level as well. Here are the relative importances I got for isolated offense, isolated defense, offensive rebounding and defensive rebounding in this column: (using 1990-91 through 2002-03 data)