Statistical analysis, charts, graphs, and observations from a lifelong NBA fan.

Permalink
[Statistical Analysis]
2005-06 Size-Adjusted Rebounding Rankings
Best and worst pound-for-pound, inch-for-inch rebounders

Nate Robinson rebounds well
for his size, too.
Last October, I developed a way of evaluating who would be the best (and worst) rebounders in the league if size weren't a factor, and posted results from the 2004-05 season. With the subject of size bubbling up as a common theme from All-Star Saturday (Dirk Nowitzki became the tallest 3 Point Shootout winner, and it really looked like Nate Robinson became the shortest Dunk Contest winner), this seems to be as good a time as any to look at 2005-06 results.

Paraphrasing my previous post:

To answer the question of who would grab the most boards if all players were sized equally, I ran linear regression analysis against NBA boxscores from the 2001-02 season through the 2004-05 season. I came up with a formula to predict rebounding results (as measured by rebounds per 40 minutes) based on height and body mass index (a measure of girth that factors out height).

Given this formula, I was able to predict each player's expected rebounding output, given his height and weight. I then compared these expected results to actual results from the 2005-06 regular season, to determine who has exceeded (or failed to meet) expectations by the widest margin. I then ranked the players by their size-adjusted rate, which is actual rebounds per 40 minutes divided by expected rebounds per 40 minutes. Average rebounders for their size achieve a size-adjusted rate of 100%. All players with at least 300 minutes played so far were included.

UPDATE: As one reader astutely pointed out, simply dividing the actual rebounds by the expected rebounds led to a bias at the top and bottom of the scale, since variation is not uniform. To correct this, I split the players into subgroups according to their size and performed separate regressions for each subgroup. For each player, the difference between his actual vs. expected rebounds is divided by his subgroup's standard deviation to calculate a new size-adjusted score. As a result of this change, there was some minor movement in each table.

Best Size-Adjusted Rebounders (through 2/16/06, min. 300 minutes)

Rank Player Height Weight Rebs/40 mins Expected Size-Adjusted Score
1. Bonzi Wells 6' 5" 210 9.21 5.15 3.87
2. Chris Paul 6' 0" 175 6.28 2.65 3.47
3. Marcus Camby 6'11" 225 14.59 8.95 2.90
4. James Singleton 6' 8" 215 11.69 6.50 2.66
5. Reggie Evans 6' 8" 245 14.01 8.95 2.60
6. T.J. Ford 5'11" 162 4.89 2.28 2.48
7. Jason Kidd 6' 4" 210 7.60 5.15 2.33
8. Ben Wallace 6' 9" 240 13.54 9.22 2.22
9. Tyson Chandler 7' 1" 235 13.32 10.01 2.14
10. Shawn Marion 6' 7" 228 11.53 7.37 2.13
11. Bobby Jackson 6' 1" 185 5.09 3.06 1.94
12. Jeff Foster 6'11" 242 13.81 10.20 1.85
13. Joe Smith 6'10" 225 11.91 8.43 1.79
14. Kevin Garnett 6'11" 220 12.21 8.84 1.73
15. Jannero Pargo 6' 1" 175 4.29 2.60 1.62


T.J. Ford

Chris Paul
Ranking #2 and #6 on the list for this year are a couple of point guards featured in Friday's Rookie Challenge game: Chris Paul and T.J. Ford. Both of these guys play a lot bigger than their size would suggest, and Paul grabs more than twice as many rebounds as he should. Two other diminutive point guards appear further down on the list: Bobby Jackson and Jannero Pargo. A couple of bigger guards, Bonzi Wells and Jason Kidd, are also faring well.

Also included among the top 15 players: a few rebound-only specialists (James Singleton, Reggie Evans, Jeff Foster), and the last 2 repeat rebounding champs (Kevin Garnett and Ben Wallace).

Worst Size-Adjusted Rebounders (through 2/16/06, min. 300 minutes)

Rank Player Height Weight Rebs/40 mins Expected Size-Adjusted Score
316. Justin Reed 6' 8" 240 4.2 8.7 (2.31)
315. Pat Garrity 6' 9" 238 4.8 9.1 (2.22)
314. Mark Blount 7' 0" 250 6.5 9.9 (2.19)
313. Clifford Robinson 6'10" 240 5.6 9.8 (2.16)
312. Brian Scalabrine 6' 9" 235 4.6 8.8 (2.14)
311. Gordan Giricek 6' 5" 210 2.9 5.2 (2.12)
310. Toni Kukoc 6'11" 235 5.7 9.6 (2.02)
309. Jason Collins 7' 0" 255 7.0 10.1 (2.00)
308. Kareem Rush 6' 6" 215 3.6 5.6 (1.93)
307. Vitaly Potapenko 6'10" 285 7.7 11.3 (1.84)
306. Matt Bonner 6'10" 240 6.3 9.8 (1.79)
305. Wally Szczerbiak 6' 7" 244 4.8 8.2 (1.75)
304. Joe Johnson 6' 7" 230 4.1 7.5 (1.73)
303. Jarron Collins 6'11" 252 7.8 11.1 (1.69)
302. Antonio Daniels 6' 4" 205 3.1 4.8 (1.69)

The most striking thing about this list is the fact that 3 of the bottom 15 players were involved in last month's big Celtics-Timberwolves trade.

Besides the notoriously board-phobic Mark Blount, the Celtics unloaded worst overall size-adjusted rebounder Justin Reed. In return, they received another bottom-15 rebounder: Wally Szczerbiak. They also managed to retain the 5th-worst sized-adjusted rebounder: Brian Scalabrine. Of the other players involved in the trade, Marcus Banks, Ricky Davis and Michael Olowokandi are also below-average rebounders for their size, and Dwayne Jones hasn't played enough to warrant a rating.


Even before the regression analysis, I knew C. Paul was going to top the list. What I was wondering more was how well D. Rodman (not so much height but weight)would do in terms of rebounding completing against guys these days who tend to be much bulkier and bigger in size like B. Wallace and D. Howard. This was actually talked a lot about during the all-star, discussing how players these days have changed so much to bigger and more powerful players.
I am almost positive that the error terms of rebounding rates will be heteroscedastic. There are only that many rebounds in a game, so the the volatility around higher predicted values will be smaller than around lower predicted value. As a result, your percentage indicator for the size adjusted rate is biased against taller guys.
Great point. I corrected this by splitting the players into subgroups according to their size and performing separate regressions for each subgroup. For each player, the difference between his actual vs. expected rebounds is divided by his subgroup's standard deviation to calculate a new size-adjusted score.

As a result of this change, Jason Kidd and Tyson Chandler moved up into the top 15, while Rafer Alston and Speedy Claxton dropped out. Martell Webster, Antonio Burks, Jalen Rose, and Marcus Banks moved out of the bottom 15, while Jason Collins, Vitaly Potapenko, Matt Bonner, and Jarron Collins slipped in.



Add a comment

Title
Body
HTML : b, i, blockquote, br, p, pre, a href="", ul, ol, li
Math Quiz 9 + 7 = (Helps stop blog spammers)
Name
E-mail address
Website
Remember me Yes  No 

E-mail addresses are not publicly displayed, so please only leave your e-mail address if you would like to be notified when new comments are added to this blog entry (you can opt-out later).