APBRmetrics

[dexterfishmore]

Hi, all. I'd like to solicit some views about how team rebounds affect the calculation of rebounding percentage. I apologize if this is covered elsewhere on the board. If it has been, feel free to direct me to the prior thread.

Here's the background to my questions....

At Silver Screen and Roll, I calculate, for every Laker game, the Lakers' and their opponents' team rebounding percentages. The easy inputs for this calculation are the rebounds credited to individual players. But team rebounds, as we all know, are a mixed bag. Some of them are meaningless accounting entries, such as when a player misses a technical foul shot or the first of two free-throw attempts. Some of them, though, reflect actual rebounding skill, such as when an opponent commits a loose-ball, over-the back foul, or when the defending team boxes out the shooting team and allows a missed shot to roll out of bounds.

In calculating team rebounding percentages, what I've been doing is going through the play-by-play data and (i) disregarding those team rebounds that don't reflect any rebounding skill on the part of the credited team, while (ii) including in the calculations those team rebounds that do reflect actual rebounding skill. The outcome, I believe, gives an accurate view of how the teams rebounded in the game, but because of my treatment of team rebounds, the resulting rebounding percentages differ from what you'd get if you looked only at individually credited rebounds.

OK. So that's what I've been doing to date. My questions are....

1. Is this standard practice? Like, when I see team rebounding rates on Basketball Reference or Basketball Prospectus, are they taking the same approach as the one I describe above?

2. If it's not standard practice, how do people typically approach team rebounds? Are they just wholly ignored?

3. Whether or not it's standard practice, is my approach analytically sound?

Thanks for reading, and thanks in advance for all replies.

[Kevin Pelton]

1. No
2. Yes
3. Yes

I think if it was more practical on a large scale, that's how everyone would handle team rebounds. I suppose it might be possible to do for play-by-play based systems, but if you're getting data at the box-score level, it's not realistic.

It might be interesting to know, if you could track it, what kind of difference team rebounds make. Are they evenly distributed or do they reflect overall team rebounding skill?

[dexterfishmore]

Thanks, Kevin. I appreciate the response.

I actually don't know the answer to the question you pose. I've got team rebounds data for all of last year's Laker games in my files, but it's all pretty disorganized - on scraps of paper lying around everywhere. In the coming season I'm going to track it more systematically, and hopefully that'll shed some light on the issue.

[mtamada]

This could be a small but solid contribution to NBA research all by itself. I wouldn't be surprised if some analysts have studied it already -- but have had to keep their results secret. But yeah, look at the numbers that you get when you ignore team rebounds, look at the numbers that you get when you include them, and compare. Which numbers show greater stability, and greater predictive power for next year's numbers? Which numbers show greater predictive power for the rebounding numbers for the next game?

If the ind rbd + team rbd numbers turn out to be better indicators of "true" rebounding ability, then it'd be another number to add to the list of things that the NBA could track and put into box scores.

One complication just occured to me: maybe the best indicator of rebounding prowess will turn out to be a weighted combination of individ and team rbds, i.e. maybe each team rebd should receive only half the weight of an individual rebd. Or maybe the weights differ for team off rebds vs team def rebds.

[greyberger]

Team offensive rebounds?

I thought almost every team rebound and uncredited rebound was defensive

[dexterfishmore]

@greyburger - There are team offensive rebounds when, say, a defensive player commits a loose-ball foul going for a rebound or a blocked shot lands out of bounds.

For instance, in this play-by-play of the fourth quarter of Lakers-Celtics Game Seven, see the entries at 11:07 and 10:11....

http://espn.go.com/nba/playbyplay?gameId=300617013&period=4

[gabefarkas]

[basketballvalue]

I do try to account for this in the numbers on basketballvalue.com. One of my things to verify this month is that i'm handling it correctly.

Thanks,
Aaron
_________________
www.basketballvalue.com
Follow on Twitter

[dexterfishmore]

Thanks, Aaron. When you say you try to account for this, are you using basically the same method I outline above - separating "earned" from "unearned" team rebounds, and disregarding the latter?

[basketballvalue]

Yes. I'm trying to filter out rebounds like the rebound listed in the pbp when someone misses their first free throw....

Thanks,
Aaron
_________________
www.basketballvalue.com
Follow on Twitter