New Rebounding Statistic

[Kirobaito]

I've been thinking about this for about five years now (since I took AP Statistics my senior year of high school).

There needs to be a new rebounding statistic measured, which I'm tentatively calling REB+ (to mimic the OPS+ statistic in baseball). Regular rebounding rates are flawed, because they are too swayed by field-goal percentage and opponents' field-goal percentage. A defensive player should get the rebound, based on positioning on the court. So, a team that shoots significantly better than their opponent should come away with more rebounds, with rebounding ability being equal. Last year, the Boston Celtics shot over three percentage points higher than their opponents (48.3-45.1), but came in behind in total rebounds. How bad were they at rebounding? Even worse than the regular statistic. A team being outrebounded doesn't necessarily make them a bad rebounding team. Maybe it's their huge field goal percentage disparity with respect to their opponents.

Some of the things we'd need to do:

1) Separate field goal misses from free-throw misses on 2nd shots/and-ones. We want to consider them separately, because 95% (guess) of the time the defense gets the rebound on a missed free throw, compared to around 70% of the time on field goals.

2) Calculate offensive rebounding percentage, defensive rebounding percentage, FT oreb percentage, and FT dreb percentage for every team.

3) Get the total percentages for the entire NBA, so we can set up a normalized system of measurement (with 100 being average, like OPS+ in baseball).

4) The last part is that some statistician needs to find an appropriate way to weigh all of these four rebounding percentages relative to one another. Maybe weigh them based on total number of rebounds for each one? So, field goal defensive rebound, field goal offensive rebound, free throw defensive rebound, free throw offensive rebound.

Our end result should be a single number that best encapsulates a basketball team's ability to rebound the ball, with 100 being average. There are probably other factors to consider that I'm forgetting, but at the very least it would do better than the current system of just using total rebounds, which is incredibly flawed.

Any thoughts? Criticisms? Does something like this already exist that I haven't noticed?

[chumdawg]

Do you think you would accomplish most of what you want to accomplish simply by using rebounding percentages?

[Kirobaito]

Originally Posted by chumdawg

Do you think you would accomplish most of what you want to accomplish simply by using rebounding percentages?

Well, basically all I'm doing is conglomerating rebounding percentages into one number. You can obviously say that a team's defensive rebounding rate is 67.8%, which ranks 24th in the NBA, and an offensive rebounding rate of 22.1%, which is 13th. Or you can say that they have an REB+ of 98, which of course slightly below average. The difficult part is separating out the different types of rebounds, which is tedious but necessary (and which you do for straight rebounding percentages anyway). After that, normalizing and weighting them is not hard or time-consuming at all. I'm looking for one number that tells me a team's overall rebounding prowess, and no single rebounding percentage can do that.

[chumdawg]

No, I get what you are trying to do. What I am asking is whether you think the stat you end up with is going to be (much) more meaningful than the RB% stats that already exist.

You might even ask whether you are making the info less meaningful when you combine offensive and defensive numbers into one singular number.

[Kirobaito]

Originally Posted by chumdawg

No, I get what you are trying to do. What I am asking is whether you think the stat you end up with is going to be (much) more meaningful than the RB% stats that already exist.

You might even ask whether you are making the info less meaningful when you combine offensive and defensive numbers into one singular number.

I'm gonna use a comparison to OPS+ in baseball, because I think it's a meaningful comparison. Both statistics are conglomerations of independent statistics (on-base percentage and slugging percentage), which are then normalized (by position, in OPS+'s case) to give a single measure of a player's offensive abilities. Is OPS+ more meaningful than just looking at OBP or SLG?

This is a combination of independent statistics, which are normalized to give a single measure of a team's overall rebounding abilities. I think there is value in that kind of a number, because I don't think it requires any other context. An REB+ of 115 will tell that this is a good rebounding team, and you don't have to know anything else to know that. Rebounding percentages are obviously the meat that make up that number, but rebounding percentages aren't very useful if you're not also given how those percentages compare with those of other teams.

[chumdawg]

Yeah, I see what you mean. That does make sense. Maybe such a thing is exemplar of the many advanced metrics the Mavs track.

[Dirkadirkastan]

Originally Posted by chumdawg

You might even ask whether you are making the info less meaningful when you combine offensive and defensive numbers into one singular number.

That is something to consider. Teams with different combinations of percentages may wind up with the same overall rating. The question would be whether it would be reasonable to say those teams do in fact demonstrate same level of rebounding strength. Obviously it would be a problem if that did not seem to be the case, but if it did, then the rating would be a useful means of summarizing and categorizing levels of overall rebounding strength.

[dirno2000]

Rebounding rate already accounts for the fact that a good defensive team has more opportunities to get defensive rebounds and a good offensive team has fewer opportunities for offensive rebounds. Sounds like the one additional thing you'd be accounting for is the fact that defensive rebounds off of free throws are easier to get then defensive rebounds off of missed field goals. Would the new ranking method be much different from total rebound rate? I guess it would depend on the weighting you assign to two rebound types (FT vs. FG).