The Difficulty of Assessing Value in Individual Rebounding

[TrueLAfan]

(a.k.a. "Why WP can't work")

Putting rebounds in a team context is one of the most difficult things to do in terms of statistical analysis. It's sort of my area. The upper limit of rebounding hampers attempts to improve team rebounding; as other posters (such as ElGee) note on other threads notes, rebounders start to cannibalize from each other pretty quickly. The rule is simple; good rebounders take boards from lesser rebounders. David Robinson was an really good rebounder. But when he played alongside Dennis Rodman--an even better rebounder--DRob's individual numbers plummeted.

So we need to think about what to make of these super-elite reounder, and how to gauge and classify them. I call these players "impact rebounders." Basically, the Reb% averages by positional starters looks something like this (starters rebound slightly better than non starters, which is why the totals add up to more than 50):

PG--5.3
SG--6.6
SF--8.8
PF/Frontcourt Player--14.3
C/Frontcourt Player--15.5

If a player rebounds at a rate about 5-6 higher than his positional averages, you have an "impact rebounder." These players are very rare. you have to assess them in funny ways. Magic Johnson is a terrific example. his career rebound% is 11.1...about 5.8 higher than a "normal" PG. A normal starting PG that plays 2900 minutes will grab about 267 rebounds. In that number of minutes, Magic will grab about 560 rebounds--nearly 300 more boards. Magic Johnson alone is the difference between being on a bad rebounding team (-150 rebound differential) and a great rebounding team (+150 differential). Or, to put it another way, Dwight Howard's Reb% of about 21.2 is more or less equal to the positional difference of Magic Johnson. They are more or less equal in team rebounding impact.

The number of impact rebounders in NBA history is very small. And impact rebounders are rarely affected by teammates. Again, good rebounders take boards from lesser rebounders. So Magic Johnson is going to have a great Reb% regardless of where he plays. So is Dwight Howard. So is Dennis Rodman.

[TrueLAfan]

Then there's a final, major component, and that's the impact of keeping opponents off the boards. If you are a good rebounder, but don't box out and allow your opponent to rebound well, the impact of your rebounding largely illusory. The best contemporary example of this is Marcus Camby. Marcus Camby is, statisically, an all-time great rebounder. His Reb% in the last decade has been 20.2--which is on the border of being Impact Rebounder level. The problem is that Marcus Camby's teams have been mediocre to poor rebounding tams. I mentioned this in a couple of months ago in another thread, and a poster questioned whether it might be teammates or injries or some other confluence of factors. The answer to that, pretty much categorically, is no. Camby has played for different teams with different lineups and different coaches. The teams had one thing in common...Marcus Camby. None of the teams was a great rebounding teams; many were poor to very poor. this season is a perfect example. Marcus Camby was traded for Steve Blake and Travis Outlaw. Adding a player of Cambv's rebounding power should have improved the Blazers rebounding. It did not. The Blazers had a slightly higher rebound differential in games without Marcus Camby.

What are the reasons for this? There are a lot of things that can affect this area.

1) Mid-post offense. If a C plays in the mid-post, the defender is near enough the basket to get rebnouncs, and Cs often have trouble boxing out in the extar space as you move father form the basket. A jump shooting C, on the other hand, can opull an opposing defencder far enough out so that it's harder to get defensvie rebounds.

2) Scoring. Part of the previous note is predicated on scoring. If an opposing defender needs to concentrate on defense and/or coming out to stay with a C that has range, it will reduce defensvie rebounding.

3) Boxing out. Pointy elbows + good position = bonuses in keeping opponents off the boards.

4) Intelligence/Skill. Some players are simply good at keeping opposing players from getting to the ball. Bill Walton was like this (a lot of Wooden/UCLA players were).

[TrueLAfan]

Well, first, and analysis that puts a heavy emphasis on individual rebounding is going to be questionable at best. Dave Berri is done like dinner. He's looking at individual rebounding numbers, and keeps saying how great a player Marcus Camby is because Marcus Camby has a very high individual Reb %. But he's missing the forest for the trees. The Blazers were 17-6 when Camby played, which is great...but 13 of those games were home games, and they played below average teams on the whole. And, as noted previously, they did not improve in rebounding. then they went to the playoffs and got bounced in 6 (and handily outrebounded) by Phoenix. Berri just couldn't have been more wrong.

And as long as he uses individual Reb% as an integral part of his WP system, he'll continue to make mistakes. Not always, and not always large. But individual rebounding numbers are not related to team rebounding in a variety of ways, which has a variety of impacts. In the case of Marcus Camby, his rebounding makes him appear more valuable than he is.

Another player who needs to be reevaluated is Kareem Abdul-Jabbar. It's sort of accepted wisdom that Kareem was a horrible rebounder after 1981. But from 1982 to 1986, the Lakers outrebounded their opponents by close to 4% on average--a huge amount. Essentially, there were no more rebounds for Kareem to get. The impact rebounding of Magic creates the illusion that Kareem became a "bad" rebounder. But the fact is that Magic cannibalized rebounds form Kareem, because good rebounders take rebounds from lesser rebounders. Kareem was, likely, still a decent rebounder (my guess is that without Magic, Kareem would have had a Reb % around 15.5 to 16 between 1982 and 1986). Kareem's years alongside Magic are similar to David Robinson's period alongside Rodman. DRob's Reb % alongside Rodman was between 14 and 15--about 10% higher than Kareem's alongside Magic. So I question whether Kareem's lower individual rebounding numbers should be statistically assessed as hurting his value greatly. His low rebounding was not in any way related to poor team rebounding--or, for that matter, to victories. The 1982-6 Lakers were one of the greatest teams of all time.

[ElGee]

So, the question is are impact rebounders at the interior positions better than impact rebounders on the perimeter??

I'll return to my original thinking on the topic -- if bigs are pulling down a larger percentage of available rebounds, there are fewer to lose as a team. Jason Kidd may well be an "impact rebounder" from the point guard position (specifically, defensive rebounding), but if he's +7.0% (relative to an avg. PG) in rebounding rate is it as good as Dwight Howard being +7.0% (relative to avg. C's)?

Going back to your example of Magic in 1982. Is he driving the Lakers to that +233 rebounding differential? How do we factor in Landsberger, Rambis, Kareem, etc.?

Also, you like playing with these formulae and percentages...so I guess the ultimate question is can we come up with a way to put a player's TRB% in context with his team's TRB% numerically??

[Idunkon1stdates]

What is Berri's response to your argument (and others like it)?

[azuresou1]

Great post TrueLAFan.

Berri is garbage.

[mysticbb]

Well written, TrueLAFan. Just want to point out that sometimes players are not playing the same position on offense as they play on defense like Magic.

Camby never really had a big impact on the rebounding, especially on the defensive end. Most of his teams were actually a better defensive rebounding team with him not on the court. Rebounding isn't just grabbing the rebound, but also boxing out and get into a position which will help your TEAM to get that rebound. Fighting your teammates for rebounds will not improve the overall rebounding.

And while Jason Kidd always had some nice rebounding numbers on the Nets, it wouldn't be possible without Collins or Martin boxing out for him. The Nets wanted the ball in Kidd's hand as soon as possible. Adding the fact that Kidd has really great instincts where the ball goes after a missed shot, he was always in a good position to either grab the board right away or receive the ball after someone tipped in his direction. You can see something similar on the Mavericks (well, Kidd doesn't even play PG on defense for them, but most times SG or even SF, which isn't included in Berri's calculation and one of the biggest reasons I think a positional adjustment is rather difficult to implement anyway), Dampier, Haywood or Nowitzki just tipping the ball into Kidd's direction and Kidd gets the credit for the rebound.

[sp6r=underrated]

What is Berri's response to your argument (and others like it)?

I'll admit part of this reason I don't like Berri is because of how arrogant he is. Berri generally responds to criticism by asking you to put your article in a peer reviewed journal.

[ElGee]

I've been playing around with the numbers and my guess is there is a correlation between the Marcus Camby Impact Illusion (MCII) and the expected rebounding differential a team should have. I've only done this for a few teams, but here's the gist of it:

*we can estimate the expected rebounding differential of an individual at his position based on is TRB%, his positional average, the % of his teams minutes he played, and the total available rebounds in his team's games. (Obviously, positional adjustments can be a bit fuzzy.)

*if that number is GREATER than the actual team differential, it (in theory) represents players who are taking rebounds away from their teammates instead of the opponent.

*if that number is SMALLER than the actual team differential, it (in theory) represents players who are taking rebounds away from their opponents instead of their teammates.

Again, I've only looked at a handful teams, but it germinates from the evidence that players who are clearly having a positive rebounding effect on team differential (eg Rodman) show this pattern in their teams, while others (eg Camby, Kemp) show the opposite pattern.

For example:
1996 Bulls w/out Rodman: +4.1 rebounding differential (18g)
1996 Bulls w/Rodman: +7.3 rebounding differential

1992 Sonics w/out Kemp: +4.6 rebounding differential (18g, Kemp's best year by TRB%)
1992 Sonics w/Kemp: +1.7 rebounding differential

If we run the method describe above on ~the top 95% of MP for the entire team and sum the results, the expected differential of the teams are:

1992 Sonics: +270
1996 Bulls: +261

The actual differential of those teams?
1992 Sonics: +189
1996 Bulls: +541

So, the team where Kemp is scooping up the majority rebounds, the 92 Sonics, has 81 fewer rebounds than we'd "expect." The 96 Bulls have 280 MORE. The idea is the leading rebounders based on differential (Rodman+441 and Jordan +191) are grabbing their boards from opponents while Kemp (+216) or Michael Cage (+120) has a MCII situation and some of their differential comes at the expense of his teammates.

This leads me to believe there is something to this, as it shows up with other teams as well where we can roughly isolate a player's rebounding effect on the entire team. I've just started to play with this, so further investigation is welcome!

[azuresou1]

Can you run the numbers for Shane Battier?

[ElGee]

Can you run the numbers for Shane Battier?

Is there a particular reason you want Battier? His numbers are average for his position and he's a non-big anyway, so it would be tricky to conclude anything as far as I can tell...

[azuresou1]

Well the consensus about him was that although his individual rebound numbers didn't pop, he made his teammates better by boxing out very well - I want to see if there's any validity to that.

[Dr Positivity]

To me there are two sides of rebounding. There is blocking out and using your body, and there is what I like to call the arms and legs portion, which is going up to get the ball. I've always had a hunch the guys who emphasize the former are more valuable than their numbers and the guys who emphasize the latter are less. I'm not saying Rasho has more impact as a rebounder than KG, but KG doesn't have 2x the impact like the numbers would indicate.

I believe the goal of rebounding should be to prevent offensive rebounds, rather than to grab defensive rebounds. This is because the damage of an offensive rebound is higher than the benefit of a defensive rebound, due to the latter happening more often. Also an offensive rebounding is usually individually based, one player on the court gets 'beat'. Whereas defensive rebounding is arguably more of a collaboration. It's not only the guy who grabs it that's important, but the other 4 guys covering their player. Thus allowing an offensive rebound is more of an individual failure than grabbing a defensive rebound is an individual benefit.

In short it's not worth it to increase the opponent's chance at an offensive rebound, to increase your chance at a defensive rebound. On the other hand if you make sure the player you're covering doesn't get it, you're contributing to the team's defensive rebounding, even if you don't get the ball.

Making a hypothetical example. There are two players. Captain Box-out and Captian Arms-n-legs. In a game Captain Box-out makes sure the opposing C never grabs a rebound, but he also never grabs any. Meanwhile Captain Arms-n-legs grabs 12 but the opposing PF grabs 4 offensive rebounds. Now on paper with our usual counting def. boards way, it'd appear Captain Box-out didn't do his job and Captain Arms-n-legs did. But I'd argue it's the opposite. Captain Arms-n-legs gave up 4 offensive rebounds and Captain Box-Out gave up 0. Those are 4 rebounding breakdowns the team can't get back. Whereas the 12 defensive boards Captain Box-Out grabbed can reasonably be made up by everyone else, presuming they also box out their players. A team full of Captain Box-Outs who don't allow offensive rebounds, would grab every rebound. A team full of Captain Arms-n-legs who are great at going up and getting it but also allow offensive rebounds, would have much worse results overall

Here's also an interesting comparison

09-10 Miami Heat
Carlos Arroyo/Mario Chalmers
Dwyane Wade/Daequon Cook
Quentin Richardson/Dorrell Wright
Michael Beasley/Udonis Haslem
JO/Joel Anthony

08-09 Toronto Raptors
Jose Calderon/Roko Ukic/Will Solomon
Anthony Parker/Jason Kapono
Jamario Moon/Joey Graham
Chris Bosh/Kris Humphries
JO/Andrea Bargnani

According to the stats the PG, SG, SF, and C positions would all rebound evenly on this teams, but the Raptors should have the advantage due to Chris Bosh's massive rebounding advantage over Michael Beasley. But the Miami Heat ranked 6th in DRB% this year and as for us? We were 30th in the league until we traded JO halfway through the year. So that's weird. Now the easy explanation is to blame Bargnani, one the worst rebounding Cs ever. But in 06-07 and 07-08 where he logged major minutes we ranked 14th and 4th in DRB%, and this was with Rasho as the 3rd C. Furthermore after the Marion trade we were the best DRB% team in the league. So I wouldn't put it all on him to submarine this. For whatever reason this lineup just couldn't rebound. I do believe some of it is the difference between Beasley and Bosh on the boards as smaller than the stats say. Beasley rebounds with his body but can't get the ball, while Bosh is the definition of an arms-n-legs rebounder. Furthermore this year JO often left his man to try and block a shot, further compounding the problem. I think there's other variables at play though. The 08-09 Raptors allowed more 3s than anyone due to our heavy dosage of perimeter help d, and Parker and Moon were often running around as chickens with their heads cut off around the perimeter and closing out on 3s, putting them in a weak boxing out position. Our rotations were sloppy all the time. The system was lacking. Meanwhile look at the Heat. Great defensive team, great fundamentals. They played within a system and knew where they'd be every possession they shot went up. They forced more midrange shots which help defensive rebounding. Their coach emphasized d and boxing out a lot more in practice. Instead of players running around the court providing lost help d, they knew where to be. JO was probably told to box out more. Furthermore with Wade providing so much offense, boxing out was a larger portion of their "role" and weight to carry on the rest of the team, because they sure weren't doing anything offensively. I think rebounding a lot of times can come down to energy and effort as well.

One stat I'd really like to see gain steam is "Opposing player rebounds". Every time an offensive rebound is gathered, the guy who was boxing him out should be listed. Same as an assist. It'd really help the rebounding stats IMO

[azuresou1]

I think it depends for different players though.

Apparently early in college Barkley sucked as a rebounder because he kept going to box out his man, and the rebounds weren't falling his way. Then his coach told him to worry less about boxing out and more about just going for the rebounds, because he was so short that opponents could go at rebounds against him even if he boxed out perfectly, but since he had great ups and timing, that he'd outjump them for the boards. That advice worked out pretty well for Barkley, who's 13th on the all time ORB% and TRB% lists and 25th on the DRB% as a 6'4 fatass playing PF.

[ElGee]

To me there are two sides of rebounding. There is blocking out and using your body, and there is what I like to call the arms and legs portion, which is going up to get the ball. I've always had a hunch the guys who emphasize the former are more valuable than their numbers and the guys who emphasize the latter are less. I'm not saying Rasho has more impact as a rebounder than KG, but KG doesn't have 2x the impact like the numbers would indicate.

I believe the goal of rebounding should be to prevent offensive rebounds, rather than to grab defensive rebounds. This is because the damage of an offensive rebound is higher than the benefit of a defensive rebound, due to the latter happening more often. Also an offensive rebounding is usually individually based, one player on the court gets 'beat'. Whereas defensive rebounding is arguably more of a collaboration. It's not only the guy who grabs it that's important, but the other 4 guys covering their player. Thus allowing an offensive rebound is more of an individual failure than grabbing a defensive rebound is an individual benefit.

In short it's not worth it to increase the opponent's chance at an offensive rebound, to increase your chance at a defensive rebound. On the other hand if you make sure the player you're covering doesn't get it, you're contributing to the team's defensive rebounding, even if you don't get the ball.

Making a hypothetical example. There are two players. Captain Box-out and Captian Arms-n-legs. In a game Captain Box-out makes sure the opposing C never grabs a rebound, but he also never grabs any. Meanwhile Captain Arms-n-legs grabs 12 but the opposing PF grabs 4 offensive rebounds. Now on paper with our usual counting def. boards way, it'd appear Captain Box-out didn't do his job and Captain Arms-n-legs did. But I'd argue it's the opposite. Captain Arms-n-legs gave up 4 offensive rebounds and Captain Box-Out gave up 0. Those are 4 rebounding breakdowns the team can't get back. Whereas the 12 defensive boards Captain Box-Out grabbed can reasonably be made up by everyone else, presuming they also box out their players. A team full of Captain Box-Outs who don't allow offensive rebounds, would grab every rebound. A team full of Captain Arms-n-legs who are great at going up and getting it but also allow offensive rebounds, would have much worse results overall

Here's also an interesting comparison

09-10 Miami Heat
Carlos Arroyo/Mario Chalmers
Dwyane Wade/Daequon Cook
Quentin Richardson/Dorrell Wright
Michael Beasley/Udonis Haslem
JO/Joel Anthony

08-09 Toronto Raptors
Jose Calderon/Roko Ukic/Will Solomon
Anthony Parker/Jason Kapono
Jamario Moon/Joey Graham
Chris Bosh/Kris Humphries
JO/Andrea Bargnani

According to the stats the PG, SG, SF, and C positions would all rebound evenly on this teams, but the Raptors should have the advantage due to Chris Bosh's massive rebounding advantage over Michael Beasley. But the Miami Heat ranked 6th in DRB% this year and as for us? We were 30th in the league until we traded JO halfway through the year. So that's weird. Now the easy explanation is to blame Bargnani, one the worst rebounding Cs ever. But in 06-07 and 07-08 where he logged major minutes we ranked 14th and 4th in DRB%, and this was with Rasho as the 3rd C. Furthermore after the Marion trade we were the best DRB% team in the league. So I wouldn't put it all on him to submarine this. For whatever reason this lineup just couldn't rebound. I do believe some of it is the difference between Beasley and Bosh on the boards as smaller than the stats say. Beasley rebounds with his body but can't get the ball, while Bosh is the definition of an arms-n-legs rebounder. Furthermore this year JO often left his man to try and block a shot, further compounding the problem. I think there's other variables at play though. The 08-09 Raptors allowed more 3s than anyone due to our heavy dosage of perimeter help d, and Parker and Moon were often running around as chickens with their heads cut off around the perimeter and closing out on 3s, putting them in a weak boxing out position. Our rotations were sloppy all the time. The system was lacking. Meanwhile look at the Heat. Great defensive team, great fundamentals. They played within a system and knew where they'd be every possession they shot went up. They forced more midrange shots which help defensive rebounding. Their coach emphasized d and boxing out a lot more in practice. Instead of players running around the court providing lost help d, they knew where to be. JO was probably told to box out more. Furthermore with Wade providing so much offense, boxing out was a larger portion of their "role" and weight to carry on the rest of the team, because they sure weren't doing anything offensively. I think rebounding a lot of times can come down to energy and effort as well.

One stat I'd really like to see gain steam is "Opposing player rebounds". Every time an offensive rebound is gathered, the guy who was boxing him out should be listed. Same as an assist. It'd really help the rebounding stats IMO

Excellent post. "Opposing Player rebounds" is the holy grail of rebounding stats, IMO. I'm not even sure if teams track this...

[jicama]

Don't worry about position averages. Just multiply Reb% by (TmReb/OpReb) for the "corrected" version.
In 2009, Camby's Reb% was 20.7 . But the Clipps got only 90% as many rebounds as their opponents.
90% of 20.7 is 18.7 -- the % of "available" rebounds Camby got.

[penbeast0]

(a.k.a. "Why WP can't work")
PG--5.3
SG--6.6
SF--8.8
PF/Frontcourt Player--14.3
C/Frontcourt Player--15.5

If a player rebounds at a rate about 5-6 higher than his positional averages, you have an "impact rebounder." These players are very rare. you have to assess them in funny ways. Magic Johnson is a terrific example. his career rebound% is 11.1...about 5.8 higher than a "normal" PG. A normal starting PG that plays 2900 minutes will grab about 267 rebounds. In that number of minutes, Magic will grab about 560 rebounds--nearly 300 more boards. Magic Johnson alone is the difference between being on a bad rebounding team (-150 rebound differential) and a great rebounding team (+150 differential). Or, to put it another way, Dwight Howard's Reb% of about 21.2 is more or less equal to the positional difference of Magic Johnson. They are more or less equal in team rebounding impact.

The number of impact rebounders in NBA history is very small. And impact rebounders are rarely affected by teammates. Again, good rebounders take boards from lesser rebounders. So Magic Johnson is going to have a great Reb% regardless of where he plays. So is Dwight Howard. So is Dennis Rodman.

Mysticbb touched on this but I want to expand a bit. Magic's rebounding (like Oscar's before him) is overrated because he is classified as a PG when he played the 2 and 3 positions defensively. Look at the rebounding numbers for Norm Nixon (who even outassisted Magic as his backcourt mate early on) and Byron Scott -- they were the PGs on that offense defensively and thus for the purpose of looking at rebounding (since defensive rebounding is more common than offensive). Same with Oscar playing next to a 6-2 small guard like Adrian Smith. If Oscar is defending the opposition's big guard, he should count as a 2 for rebounding purposes.

Magic still shows as an impact rebounder as a 2, just not running completely away from the pack -- and Jason Kidd, or someone like him who actually defends mainly 1s moves up in the PG charts to #1.