Duke's Defensive Rebounding: Numbers That Matter

[tommy]

OK so the big concern that a lot of folks had coming into this season was Duke’s ability (or inability) to defend against big post players, and our ability (or inability) to rebound off the defensive glass. Many were worried about this because of Duke’s lack of size at center, with the slender 6’9” Amile Jefferson the starter in the middle, and only the not-ready-for-prime-time Marshall Plumlee on the bench and over 6’8” tall.

I am attempting to track, using the numbers, just how well or how poorly we’re doing in these areas. I’m going to do it in two threads, one for defensive rebounding and one for interior defense. These threads may not be for everyone, but for those who do like to drill down a bit, or see what others find by drilling down, it may yield some interesting and perhaps surprising results. I do want to say up front that I have been very fortunate to have Kedsy’s help in coming up with many interesting ideas as to how to format the raw data I was recording as well as suggestions of many different ways to look at the data. This analysis and these threads would not have happened without him. Thank you Keds!

So: methodology. This is the rebounding thread, so I’ll only cover that analysis here. See the companion thread on the interior defense for the more complicated analysis of that aspect of the game.

The defensive rebounding analysis is very simple. When I chart the game, I am first of all taking long rebounds, rebounds that go out of bounds, rebounds that are essentially loose balls and scrums, and other miscellaneous stuff like that out of the mix. Why? Because who gets those balls has little or nothing to do with the issue being studied – the impact of our lack of height and bulk down low on our ability to rebound. The official box scores always include higher numbers of offensive rebounds than I record, because they count these miscellaneous plays as rebounds. I detailed a number of these from the Kansas game, for example, in the Kansas game thread, pointing out how the Jayhawks’ official offensive rebounding numbers were deceivingly high because of a number of these “non-traditional” shall we say, rebounds, and rebounds that had nothing whatsoever to do with size. Duke blocks a guy’s outside jumpshot and it results in a loose ball flying in the air and landing in a Kansas guy’s hands on the perimeter and he gets credit for an offensive rebound. I’m not counting that kind of thing.

I’m recording the number of “short” defensive rebounds that we get. And then I’m looking at the opponent’s “short” offensive rebounds, and further sub-categorizing those into short offensive rebounds that lead to quick putbacks or other short shots, those that lead to an open 3-pointer, and those that lead to re-setting the possession.

Then, the two ratios I am calculating are our short defensive rebounds over the sum of our short defensive rebounds plus their short offensive rebounds of all types – let’s call that the “Short Rebounding %”, and then, slightly different, our short defensive rebounds over the sum of our short defensive rebounds plus their short offensive rebounds but not counting those that lead to re-set possessions, as those are less damaging than the other types of offensive rebounds, or at least less immediately damaging. Let’s call that the “Modified Short Rebounding %.” The higher the number, the better we’re doing on the defensive boards.

Ok so the rebounding numbers so far. As you can see, on short rebounds, the ones that many expected us to struggle to grab against big teams, we actually have fared the best against perceived behemoth Kansas, while the other three, smaller opponents that I’ve charted so far have grabbed a higher percentage of those types of rebounds against us than did the Jayhawks. One tentative conclusion that can be drawn, from this obviously very small sample size, is that when we are getting beaten for these types of rebounds, it’s not because the opponents’ big guys are taller than ours.

Also, when Kansas’ bigs did beat us for an OR, they more frequently had to just pass it back out for a re-set, as opposed to doing real immediate damage in terms of scoring it. Hence, the bigger gap between the two percentages in the Kansas game than in the other two games.




Short Rebounding Raw Numbers
Short Rebounding %
Modified Short Rebounding Raw Numbers
Modified Short Rebounding %


Kansas
12/18
.667
12/15
.800


Davidson
21/32
.636
21/31
.677


UNC-Asheville
19/30
.633
19/28
.679


ECU
18/28
.643
18/26
.692










I’m very well aware that this is a very small sample size. There are also, I’m sure, multiple ways to look at data like this and multiple conclusions that reasonable people can draw. The same is true, even more so, with regard to the defensive numbers in the other thread. Just putting it out there for discussion, hoping to shed a little light, that’s all. I’ll be updating best as I can (though Vermont may be a tough one due to travel) so long as you guys are interested in it.

[CDu]

I know you put a lot of work into this, but there is a big problem with this analysis. Interior defense and defensive rebounding are highly correlated, so to understand one result you need to also understand the other.

Additionally, I think your methodology of excluding balls that go out of bounds and balls that hit the floor is a flawed approach. I would argue that a smaller/less rebounding-oriented team would be more susceptible to these events by virtue of not having the size/positioning to prevent more of these oddball rebounds. A team that lacks a big, physical rebounding presence is more likely to get pushed under the glass and out of position to field balls that don't follow the typical rebound trajectory. There are certainly some rebounds that truly are a crapshoot, but I'd argue that being a poor rebounding team results in more of these oddballs than you'd see from a good rebounding team. So by excluding these rebounds, you're omitting part of our weakness.

That being said, the fact that we haven't topped 70% short rebounding percentage in any game (even ignoring any potential positive biasing of the data) so far is a HUGE concern.

[Skitzle]

I'd love to see a Points off Offensive rebounds (short or otherwise) stat as well or something similar. Here's another thing that can skew offensive rebounding stats. Big man shoots, misses, rebounds shot, misses, rebounds puts it back.

I feel like there are a few of these types of plays every game...

[CDu]

I'd love to see a Points off Offensive rebounds (short or otherwise) stat as well or something similar. Here's another thing that can skew offensive rebounding stats. Big man shoots, misses, rebounds shot, misses, rebounds puts it back.

I feel like there are a few of these types of plays every game...

Sure, although no matter how you cut it, we've been bad on the glass. We've been outrebounded outright in 3 of 5 games. We've been outrebounded percentage-wise in 4 of 5 games. And 4 of our 5 opponents are actually smaller than we are. The one time we've played a bigger team, we lost the rebound battle by 15.

[tommy]

Additionally, I think your methodology of excluding balls that go out of bounds and balls that hit the floor is a flawed approach. I would argue that a smaller/less rebounding-oriented team would be more susceptible to these events by virtue of not having the size/positioning to prevent more of these oddball rebounds. A team that lacks a big, physical rebounding presence is more likely to get pushed under the glass and out of position to field balls that don't follow the typical rebound trajectory. There are certainly some rebounds that truly are a crapshoot, but I'd argue that being a poor rebounding team results in more of these oddballs than you'd see from a good rebounding team. So by excluding these rebounds, you're omitting part of our weakness.

Not really. Let's look at the Kansas game as an example. They were credited with 9 offensive rebounds in that game per the ESPN box score. On at least three, and depending on your perspective, maybe four of those nine, they had nothing to do with rebounding, size, bulk, individual rebounding skills, with what makes a team a good or bad rebounding team, or anything of the sort. To wit:


Late in the first half, Joel Embiid got his shot blocked. He ended up on the seat of his pants. The ball ends up landing in his lap, and he gets credited for an offensive rebound. I don’t think that has anything to do with his being bigger, stronger, a better rebounder, or anything of the kind.
Almost at the end of the first half, Perry Ellis recovered a loose ball also while on the seat of his pants, and got credited with an offensive rebound. Again, nothing to do with rebounding, at least in my mind.
Midway through the second half, Embiid found himself standing under the hoop at the end of a Kansas fast break. Kansas missed the layup, and because Embiid had beaten the Duke bigs down the floor, he caught the missed shot, again right under the hoop, and passed it back outside. Not only was there no “harm” really from his getting that ball, but it was a ball that bounced right to him with nobody around, that was the result of his being down the court first, not outsizing or outfighting our guys for a rebound. Sure, it’s legitimate in the official stats to credit him with an OR, but it doesn’t constitute the type of play most folks around here were concerned about coming into the season when talking about our ability to keep opposing bigs off the offensive boards.
There was another play midway through the first half where Wiggins got credit for an OR, but it was really kind of a long rebound/loose ball type of play. That’s a borderline one.

These Kansas OR’s – 3 or even 4 out of the 9 they were credited with, were not related to their big guys’ height, bulk, positioning, rebounding skill, or those kinds of attributes. That’s a significant percentage of their OR’s that are kind of phony, at least in my mind.


That being said, the fact that we haven't topped 70% short rebounding percentage in any game (even ignoring any potential positive biasing of the data) so far is a HUGE concern.

I tend to agree, but it would be interesting to see the short rebound percentages for teams that have the reputation as strong rebounding teams. They may in fact be better than 70%. Or maybe not.


Sure, although no matter how you cut it, we've been bad on the glass. We've been outrebounded outright in 3 of 5 games. We've been outrebounded percentage-wise in 4 of 5 games. And 4 of our 5 opponents are actually smaller than we are. The one time we've played a bigger team, we lost the rebound battle by 15.

Kansas did outrebound us by 15 on the official sheet, you’re right. But they got 7 defensive rebounds on missed Duke free throws, which are gimme rebounds. On the other side, Duke only got 3 defensive boards on missed KU free throws. So that’s a net +4 for Kansas right there on rebounds that Muggsy Bogues could have corralled. If you were to then subtract for some of the plays I outlined above that really had little if anything to do with rebounding/size/bulk/skill, and therefore take away another 3 or 4 from Kansas’ meaningful total, the gap starts to shrink signficantly. (I recognize I did not chart the number of Duke’s OR’s that were those oddball type of plays, so those would have to be added back, if there were any.)

Look, I’m not arguing that Duke is a great rebounding team or that it’s not an area of concern. It is. I’m saying that the raw numbers from a box score do not tell the full story of what actually is going on on the court in this regard, and that – especially in light of the fact that we have been involved in several blowouts, which tend to skew numbers – it is too early in the season to be making broad and definitive pronouncements about this or many other aspects of the team.

[DevilFalcon]

I would like to see a rebounding analysis that took shot % into account. Teams are going to have lower defensive rebound numbers against a guy like Rodney Hood because off his not missing shots. Lots of bricks make for lots of rebounds. And how do you account for steals? You can't get a rebound if the shot was never taken...

[Kedsy]

I would like to see a rebounding analysis that took shot % into account. Teams are going to have lower defensive rebound numbers against a guy like Rodney Hood because off his not missing shots. Lots of bricks make for lots of rebounds. And how do you account for steals? You can't get a rebound if the shot was never taken...

Defensive rebound percentage is the % of available defensive rebounds corralled by a player or team. Thus, shooting percentage and turnovers are normalized. Right now, Duke's DR% is 66.7% for the season. Here's the game-by-game breakdown:

Davidson: 64.9%
Kansas: 67.9%
Fla Atl: 66.7%
UNCA: 69%
ECU: 64.9%

Oddly, the Kansas game was our 2nd best defensive rebounding performance of the season (so far).

Also, the above numbers are all rebounds, not just the short rebounds that Tommy has separated out for us.

[tommy]

Updating the defensive rebounding table to add in the Alabama and Arizona games, here's what we've got:



Short Rebounding Short Rebound % Modified Raw Numbers Modified Rebound%
Raw Numbers

Kansas 12/18 66.7% 12/15 80.0%
Davidson 21/32 63.6% 21/31 67.7%
UNC-A 19/30 63.3% 19/28 67.9%
ECU 18/28 64.3% 18/26 69.2%
Alabama 18/27 66.7% 18/24 75.0%
Arizona 10/19 52.6% 10/15 66.7%


Obviously, the number that sticks out is the short rebound % against Arizona. When you include the short offensive rebounds that led to Arizona's resetting the possession, we barely got more than half the available short defensive boards (excluding the long rebounds, balls that went out of bounds, jump balls, and other miscellaneous plays not reflective of rebounding prowess, height, bulk, etc.) Nobody needs the numbers to see that Arizona has a number of long, mobile, athletic guys, and it predictably took its toll.

The only thing positive to take from this is the modified rebound percentage. This takes out the short offensive boards that lead to re-setting the possession, as these, while definitely negative plays for Duke, are less damaging, or less potentially damaging, than a short offensive board that leads directly to a put-back or tip-in or other easy offensive opportunity for the opponent. By this measure, we didn't do a whole lot worse against Arizona than we've done in previous games. Kind of makes the Kansas numbers look pretty good by comparison.

[tommy]

Adding in the Michigan numbers, here's now what we've got:




Short Rebounding Short Rebound % Modified Raw Numbers Modified Rebound%
Raw Numbers

Kansas 12/18 66.7% 12/15 80.0%
Davidson 21/32 63.6% 21/31 67.7%
UNC-A 19/30 63.3% 19/28 67.9%
ECU 18/28 64.3% 18/26 69.2%
Alabama 18/27 66.7% 18/24 75.0%
Arizona 10/19 52.6% 10/15 66.7%
Michigan 20/25 80.0% 20/25 80.0%

By the eyeball test, and by the numbers, the Michigan game was our best rebounding effort of the season. I saw us box out very strongly on a number of occasions -- it really seemed like a focus going into the game. Amile in particular caught my eye on that front. Not only that, but Michigan's offensive rebounding numbers got inflated by the last two minutes of garbage time, where McGary and I think a couple of other guys got some boards and putbacks that were essentially uncontested. So the numbers would've even been better without that. Michigan is not a great rebounding team, but still, they basically got nothing on the offensive boards against us that mattered.

[tommy]

Adding in the UCLA and Eastern Michigan games, here are the numbers. The Eastern Michigan numbers got reduced a bit by some offensive boards we gave up after the game was in hand, but still, the numbers were pretty consistent with what we've been doing all year, which is doing pretty well on the defensive boards, not great and rarely dominant, but not getting killed there either.









Short Rebounding
Short Rebound %
Modified Raw Numbers
Modified Rebound %






Raw Numbers




































Kansas

12/18


66.7%


12/15


80.0%



Davidson

21/32


63.6%


21/31


67.7%



UNC-A

19/30


63.3%


19/28


67.9%



ECU

18/28


64.3%


18/26


69.2%



Alabama

18/27


66.7%


18/24


75.0%



Arizona

10/19


52.6%


10/15


66.7%



Michigan

20/25


80.0%


20/25


80.0%



UCLA

24/33


72.7%


24/31


77.4%



East Mich.

18/27


66.7%


18/26


69.2%

[Kedsy]

Adding in the UCLA and Eastern Michigan games, here are the numbers. The Eastern Michigan numbers got reduced a bit by some offensive boards we gave up after the game was in hand, but still, the numbers were pretty consistent with what we've been doing all year, which is doing pretty well on the defensive boards, not great and rarely dominant, but not getting killed there either.

Actually, as I wrote in the other thread, our defensive rebounding is the best Duke has done in at least 18 years (I don't have numbers before 1996-97). In unadjusted overall defensive rebounding percentage, Duke is ranked #88 in the country. So right now, I think we have to say Duke's defensive rebounding is pretty darn good.

[Newton_14]

Adding in the Michigan numbers, here's now what we've got:




Short Rebounding Short Rebound % Modified Raw Numbers Modified Rebound%
Raw Numbers

Kansas 12/18 66.7% 12/15 80.0%
Davidson 21/32 63.6% 21/31 67.7%
UNC-A 19/30 63.3% 19/28 67.9%
ECU 18/28 64.3% 18/26 69.2%
Alabama 18/27 66.7% 18/24 75.0%
Arizona 10/19 52.6% 10/15 66.7%
Michigan 20/25 80.0% 20/25 80.0%

By the eyeball test, and by the numbers, the Michigan game was our best rebounding effort of the season. I saw us box out very strongly on a number of occasions -- it really seemed like a focus going into the game. Amile in particular caught my eye on that front. Not only that, but Michigan's offensive rebounding numbers got inflated by the last two minutes of garbage time, where McGary and I think a couple of other guys got some boards and putbacks that were essentially uncontested. So the numbers would've even been better without that. Michigan is not a great rebounding team, but still, they basically got nothing on the offensive boards against us that mattered.
Josh Hairston also had several good blockouts in this game.


Adding in the UCLA and Eastern Michigan games, here are the numbers. The Eastern Michigan numbers got reduced a bit by some offensive boards we gave up after the game was in hand, but still, the numbers were pretty consistent with what we've been doing all year, which is doing pretty well on the defensive boards, not great and rarely dominant, but not getting killed there either.










Short Rebounding

Short Rebound %

Modified Raw Numbers

Modified Rebound %








Raw Numbers













































Kansas



12/18



66.7%



12/15



80.0%



Davidson



21/32



63.6%



21/31



67.7%



UNC-A



19/30



63.3%



19/28



67.9%



ECU



18/28



64.3%



18/26



69.2%



Alabama



18/27



66.7%



18/24



75.0%



Arizona



10/19



52.6%



10/15



66.7%



Michigan



20/25



80.0%



20/25



80.0%



UCLA



24/33



72.7%



24/31



77.4%



East Mich.



18/27



66.7%



18/26



69.2%






As Kedsy indicated, rebounding is currently a strength for this team, when compared to prior Duke teams. Amile is now leading the charge. He had one incredible rebound Saturday where he grabbed a board up high in heavy traffic, cupping it with one hand, and securing it as he came down. It was one of those "wow, how the heck did he do that" moments.

Thanks for putting the data together. Nice work.

[-jk]

Actually, as I wrote in the other thread, our defensive rebounding is the best Duke has done in at least 18 years (I don't have numbers before 1996-97). In unadjusted overall defensive rebounding percentage, Duke is ranked #88 in the country. So right now, I think we have to say Duke's defensive rebounding is pretty darn good.

I'll take it!

It's always hard to compare fall-only games to a complete season, though. Without any evidence whatsoever, I would think both the SOS and opponents' familiarity with our talents would go up as we move into the ACC regular season and then into post-season, slowly pushing the season's numbers down. Is there any way to compare where we are now to Jan 1 numbers from older seasons?

-jk

[Kedsy]

I'll take it!

It's always hard to compare fall-only games to a complete season, though. Without any evidence whatsoever, I would think both the SOS and opponents' familiarity with our talents would go up as we move into the ACC regular season and then into post-season, slowly pushing the season's numbers down. Is there any way to compare where we are now to Jan 1 numbers from older seasons?

-jk

I imagine it might be possible with a game by game analysis, but that's way too much work for me. I suppose it could also possibly be done using web.archive.org, which would be easier but still probably beyond what I'm willing to do in the next couple weeks.

[Saratoga2]

Josh Hairston also had several good blockouts in this game.



As Kedsy indicated, rebounding is currently a strength for this team, when compared to prior Duke teams. Amile is now leading the charge. He had one incredible rebound Saturday where he grabbed a board up high in heavy traffic, cupping it with one hand, and securing it as he came down. It was one of those "wow, how the heck did he do that" moments.

Thanks for putting the data together. Nice work.

Am I imagining it or has Amile gained a few pounds since the beginning of the season? Amile has good size and very good length to go along with what appears to be a knack for getting rebounds. Add a little more bulk and strength by March and he could become even a more important asset for the team.

[tommy]

Am I imagining it or has Amile gained a few pounds since the beginning of the season? Amile has good size and very good length to go along with what appears to be a knack for getting rebounds. Add a little more bulk and strength by March and he could become even a more important asset for the team.

I noticed that too, when the camera zooms in on him when he's shooting free throws. He appears to be more muscular in his upper body.

[JayBean]

I noticed that too, when the camera zooms in on him when he's shooting free throws. He appears to be more muscular in his upper body.

Amile gained weight in the upper body since the start of the season? Me too! (The gut is part of the upper-body, right?)

[jv001]

Amile has a knack for getting rebounds. I mean for a guy that does not have great hops, he sure can get the rebounds. I guess it's a matter of blocking out well and having strong hands. He doesn't get stripped once he has the ball in his hands. Plus he looks to have basketball smarts. Good kid and a key to Duke having a great season over just a good season. GoDuke!

[tommy]

Adding in the Notre Dame numbers, here is the updated chart. Note that this is the worst we've done in the "short rebounding" department all year, other than against Arizona. Short rebounds are the ones that matter the most, IMO, as they tell us a lot more about our ability to handle teams that are tall or that are aggressive on the offensive boards. Long rebounds and balls that bounce out of bounds and "team rebounds" and scrum plays on the floor don't tell you much, and I suspect they even out in any event.

I for one was not surprised that we struggled on the boards against the Irish. In addition to the 6'11" and beefy Sherman, they also brought Auguste, Knight, and Burgett to the boards, all of whom are big boys, and Connaughton is an excellent rebounder for his size as well.









Short Rebounding
Short Rebound %
Modified Raw Numbers
Modified Rebound %






Raw Numbers




































Kansas

12/18


66.7%


12/15


80.0%



Davidson

21/32


63.6%


21/31


67.7%



UNC-A

19/30


63.3%


19/28


67.9%



ECU

18/28


64.3%


18/26


69.2%



Alabama

18/27


66.7%


18/24


75.0%



Arizona

10/19


52.6%


10/15


66.7%



Michigan

20/25


80.0%


20/25


80.0%



UCLA

24/33


72.7%


24/31


77.4%



East Mich.

18/27


66.7%


18/26


69.2%



Notre Dame

16/27


59.3%


16/24


66.7%

[tommy]

Here are the updated rebounding charts, post-Georgia Tech. Nobody should be surprised by our outstanding numbers on the defensive boards, the best all season actually. It was a very long time before the Jackets corralled even their first defensive rebound. And two of their legitimate o-boards came essentially back to back, on the same play. Yes, it helped -- a lot -- that Robert Carter, Jr. was not in the lineup for Georgia Tech, but still. Can't argue with the results. Outstanding performance in this area, in my opinion, and the numbers appear to back it up.











Short Rebounding
Short Rebound %
Modified Raw Numbers
Modified Rebound %






Raw Numbers




































Kansas

12/18


66.7%


12/15


80.0%



Davidson

21/32


63.6%


21/31


67.7%



UNC-A

19/30


63.3%


19/28


67.9%



ECU

18/28


64.3%


18/26


69.2%



Alabama

18/27


66.7%


18/24


75.0%



Arizona

10/19


52.6%


10/15


66.7%



Michigan

20/25


80.0%


20/25


80.0%



UCLA

24/33


72.7%


24/31


77.4%



East Mich.

18/27


66.7%


18/26


69.2%



Notre Dame

16/27


59.3%


16/24


66.7%



G Tech

22/27


81.4%


22/26


84.6%