NBA HD: Offensive Orientation and Re-Estimating ORR

Carl Landry came up in a conversation I had recently.

In his rookie season in 2007-08, Landry grabbed 15.5 percent of the Rocket’s missed shots while he was on the floor, a rate that placed him atop the list of every player in the NBA.  Remarkably, the Rockets found a guy in the second round of the draft who immediately stepped in as the best mess cleaner in the NBA.  Not three years down the road. Immediately.

And two seasons later, Landry is a below-average offensive rebounder at his position.

As surprising as it might be, Landry isn’t a vacuum around his basket anymore.  His offensive rebounding rate (ORR) could be micro-graphically illustrated by the backslash on the keyboard in front of you.  From 15.5 percent in 2007-08, his ORR slid down to 10.3 percent in his sophomore campaign, and 8.5 percent this past season.  But if we focus solely on his brief stint for Sacramento, his ORR stood at 7.1.  If you’re scoring at home, you’ll notice that his once elite offensive rebound production has been slashed in half.

One can look at these numbers and arrive at several conclusions.  ”Carl Landry can’t rebound anymore.“… ” Carl Landry is a phony.“… “Please, stop talking about this new-fangled stat, ORR. You sound like a damn fool.” … “OK, so what’s your point?”

Well, I think we’re missing an important piece of the puzzle: Carl Landry isn’t playing around the basket offensively anymore.   Sacramento has the Purdue alum playing further away from the rack and as a result, he has shot long twos twice as often as he did in Houston.  So he may not grab as many offensive rebounds but that doesn’t necessarily mean he’s no longer skilled as an offensive rebounder. This is an important distinction.

I bring up the case of Carl Landry because I was looking at the great work of Terrance Laney (@bbstats on Twitter) and I noticed something interesting about the players who ranked at the bottom of his position and height adjusted total rebound rate list (in laymen’s terms, who rebounds better or worse than you’d expect given their height and position).  In the bottom 10, you find players like Dirk Nowitzki, Andrea Bargnani, Channing Frye, Zydrunas Ilgauskas, Rashard Lewis and Matt Bonner.  Notice a common bond?  That’s right, perimeter shooters who happen to be very tall.

So are these tall marksmen the least skilled rebounders in the NBA or has their shooting repertoire pushed them away from the rebounding zone?  I think it’s more of the latter.  In some cases, ORR isn’t the best proxy for rebounding ability in the same way that field goal percentage isn’t as good of a proxy for shooting efficiency as effective field goal percentage for 3-point shooters.

With that in mind, I went back to the drawing board with the help of the indispensable Ed Kupfer and modeled a new expected ORR using a logistic regression that adjusts for position (PG, C, PF, SF, SG, F), height( in inches) and percentage of shots on the perimeter (PerPct) which is simply the percentage of shots outside 16 feet.   Why do we add the perimeter factor? We’re looking to isolate skill from the factors unrelated to offensive rebounding ability so this accounts for the perimeter shooters who don’t grab offensive boards because of their outside offensive orientation.  For example, Carl Landry could still be a worthy offensive rebounder even though he has migrated to the perimeter.  The model can be seen here for those who care to see the details.

So let’s take a look at the player’s who went above and beyond what we would expect given their position, height, and their shooting orientation.

Without context, a 14.2 ORR is extraordinarily impressive but after you consider that Jordan Hill took almost half his shots from the perimeter, he looks even better.  As a Knick, Jordan Hill collected tons of offensive rebounds and he might have been even more underrated than his 14.2 ORR would suggest, given his orientation on the Knicks offense.  Some might see Jordan Hill as a deal throw-in but he possesses a knack for offensive rebounding not unlike his Rockets predecessor Carl Landry.

Coming up on his 36th birthday, Antonio McDyess can still bring it.  He’s an above-average offensive rebounder even for his position and as the 8th most frequent long two taker in the NBA, his offensive rebounding rate in San Antonio is absolutely superb.  All things considered, we would expect his offensive rebounding rate to be half of what it actually is.  Impressive.

Interestingly enough, Zydrunas Ilgauskas finds himself as one of the leaders here despite ranking poorly as a rebounder in Laney’s study.  Why? Like McDyess, he’s a perimeter shooter disguised as a big man.  He cleans up the glass much better than this model would expect.

And how about the trailers?  The players below don’t rebound nearly as much as this model thinks they should.

Etan Thomas might be the Thunder’s highest paid player but this presents another reason why he has barely gotten any run recently.  He hugs the basket on offense and yet he doesn’t offer much in the way of getting second chance opportunities.  Having taken just 12 shots outside 10 feet his year, we would expect Thomas to have more boards with his rim-centric habitat.

Whereas Ilgauskas shines in this light, Dirk Nowitzki still rates as a poor rebounder on the offensive end.  80 offensive rebounds in 3,039 minutes still carries a stench even with the perimeter game adjustment that lowers his expected rate from 9.6 percent to 6.6 percent.  This is his lowest ORR since his 2002-03 season in which he, not coincidentally, posted his highest 3-point frequency of his career.

Whenever we engineer new basketball stats, we try to strip away the luck and circumstance that blemishes the current ones.  If we want to isolate the offensive rebounding skill, it helps to find out who hangs around the rim and still can’t get a board, not just those who are tall.  Perimeter shooting percentage won’t perfectly estimate a player’s offensive orientation but until we put microchips in every player’s shoe, there will always be a margin of error.  Still, this is a big step in that direction.  If you’d like to see the whole dataset, I’ve uploaded it here on Google Docs.  I’d also like to salute Ed and Terrance again for their inspiration and assistance. Cheers.

Loading image

Click anywhere to cancel