Tuesday, December 2, 2008

Re-Imagining Defensive Rebounding (plus some bonus thoughts)

A few early bonus thoughts:

While watching the Kings-Jazz game tonight, I heard Kings color-commentator Grant Napier refer to Spencer Hawes and Brad Miller as the "Ivory Towers." I did some research and found that Hawes was the one who originally came up with the term, which sounds about right. Hawes is a solid player who's improving every time I see him, but his sound bites are as goofy as ever. 

Also: Caught this at a fun blog, thought you might enjoy it. Click the link for more.

Also, a random question: does anyone know how to use greasemonkey scripts with google chrome? I'm sort of tired of having to open up separate browsers, and I like surfing with chrome but all of my scripts only work in Firefox. 

Ok, on to the main point. The last two nights, there were two very close games in which the game turned on a timely offensive rebound. The big story at the end of yesterday's Heat-Warriors game was Beasley's steal in overtime, but the game never would have gotten to that point if Udonis Haslem hadn't grabbed an offensive rebound and scored at the end of regulation. Then tonight, the 8-8 Pacers beat the 14-1 Lakers on a last second tip-in by Troy Murphy. In both cases, the moments were fitting, since Miami retrieved 45.7% of their missed shots and Indiana retrieved 36.5% of theirs (on the season, the Lakers only allow their opponents to grab 25.1% of their misses). So I've had offensive rebounding on the brain. But not in the sense in which it was discussed before, as a boon to offensive efficiency. Instead, I am wondering about how well a player helps his team prevent opponents from getting offensive rebounds.

Here's what our stats currently measure: At the team level, we have defensive rebound percentage, which tells us exactly what we want to know -- how well a team stops its opponents from getting second chance opportunities. However, we have no good way of parlaying that knowledge down to the individual level. Instead, we have individual raw defensive rebounding numbers (per game, or per 36 or per 48 minutes), and we have individual defensive rebound percentage, which tells us roughly what portion of an opponent's misses a player rebounds himself. The problem is that defensively, the goal isn't to get a defensive rebound, it is to stop the offense from getting an offensive rebound. 

Those two goal statements may sound the same, and at the team level, they are. However, at the individual level, stopping an offensive team from getting a rebound involves both going after defensive rebounds and not allowing the opponent to be in a position to rebound, and it's this latter part of the equation that isn't really measured. There's a bit of an analogy here to shot-defense, another important aspect of the game that isn't measured by any traditional stats. A big component of stopping an opponent from scoring is challenging shots and forcing low-percentage shots, but at the individual level there isn't any measure of how a player is defending opponents' shots. Seen through this analogy, defensive rebounds become as blunt of an instrument in measuring individual defense as blocked shots. (For both defensive shooting percentages and team defensive rebounding percentages, 82games.com does a good job of showing a player's effect in terms of on-court vs. off-court, but there are a number of limitations to that approach, particularly the inability to extract the impact of an individual player versus the effect of the lineups he's most likely to be a part of. 82games also does a good job of showing player and player-counterpart production, but doesn't separate offensive and defensive rebounds).

So, some preliminary thoughts that will hopefully inform the discussion:

1) When I read it, I found this study very enlightening. At the same time, I thought it supported an intuition I had anyways -- that, basically, a lot of times if a player doesn't get a particular defensive rebound, one of his teammates might still end up with it; whereas with offensive rebounds, that isn't usually the case. That intuition is presented in terms of the marginal values of individual offensive and defensive rebounds, but the gist (for this discussion) is: individual defensive rebound rates aren't really additive -- putting two prolific defensive rebounders next to each other (for instance, Chris Kaman and Marcus Camby on the Clippers this year) doesn't guarantee a great defensive rebounding team, since these rebounders are often just taking rebounds away from each other and their other teammates, as opposed to taking them away from the opposition (the study hints that on the offensive end, this isn't the case). 

2) After the Ron Artest trade, I wondered about whether Houston would be able to remain one of the better defensive rebounding teams in the league, given Artest's historically lower rebound rates. However, so far this year the Rockets are 3rd in the league in defensive rebounding. 

3) In a recent post on Nene, I discussed the fact that Denver is performing just as well this year in terms of defensive rebounding as they did last year, despite the fact that Nene has a much lower defensive rebounding rate than Marcus Camby did last year. (2) and (3) seem to provide examples of the point I'm trying to make -- that an individual can contribute to a defensive rebound without actually getting credit for that rebound, and that this contribution isn't really measured.

4) I have a feeling that defensive rebounds by point guards might be more valuable in terms of stopping the opposing team from getting a second chance than rebounds from other positions, and this link seems to agree, a little bit.

Ok. So, it's easy when watching a game to get an idea for who's doing a decent job of boxing out and who isn't, but there aren't great statistics to get a feel for that kind of thing at a summary level or for games you can't watch, and there won't be until defensive box-outs are kept as official box-score statistics. So instead, we're left guessing. For instance, I didn't get to watch tonight's Lakers-Pacers game, but I saw the recaps and the boxscores and the four factors information and the popcornmachine recap afterwards. What I gathered (other than the fact that I was right to worry about T.J. Ford) was that Troy Murphy and Rasho Nesterovic were primarily creditable for the Pacers' impressive offensive rebounding showing against the normally stingy Lakers defense. But for the Lakers, which players were responsible for the collapse? Murphy and Nesterovic combined for 4 offensive rebounds in the first half of the third quarter, when they were presumably being guarded by Pau Gasol and Andrew Bynum. However, Nesterovic grabbed four offensive rebounds by himself in the last 5 minutes of the fourth quarter (during which time the Pacers went on a 14-4 run), while Bynum was on the bench and Lamar Odom was in the game -- I'm assuming that at this point Nesterovic was Gasol's responsibility (during the same stretch Murphy, who was being guarded by Odom at that point, didn't get a single offensive rebound until the game-winner). Combined with the fact that Gasol himself only got 5 defensive rebounds in 36 minutes (and the fact that Odom was a +10 for the game while Gasol was a -6), I can sort of assume that he was the one getting beat the most often (while still acknowledging that defensive rebounding is a team responsibility). But as you can see, the process is roundabout and filled with guesswork. 

There's a number of cool stats that I've always thought could and should be part of normal boxscores. For instance, FGM-against, FGA-against, and points-against, which would track the efficiency of shots when a particular player was the closest defensive player to the shooter as well as the number of times a player was challenged during a game (here's an example of Kevin Pelton charting this type of data). Another cool stat would be a completion percentage on post-entry passes, which could add another layer of understanding to the passing game on top of assists. And now, I'm thinking I should add defensive box outs to that list, maybe with a box-out success ratio (how often a player's box-out stopped the boxed out player from making the rebound). 

1 comment:

  1. So I got a chance to watch the Pacers-Lakers highlights, and figured I should make a semi-correction. Bynum and Gasol were the two bigs in for the last play (Odom came out of the game), and the lapse wasn't Gasol's fault. At the start of the play, when Ford had the ball, Kobe lost Marquis Daniels on the weakside, and Daniels ended up with the ball and a wide open layup attempt. Gasol rotated over to challenge Daniels' shot, and Kobe should have picked up Murphy when he came in for the rebound.