View Full Version : Looking a little closer at our Defensive Rebounding
tommy
11-30-2012, 12:16 PM
CDu, thanks for this post in the Phase I thread on the issue of defensive rebounding/opponents offensive boards:
I did a quick review of the ESPN box scores for each game. Here is the breakdown:
GSU: 7 for the forwards and centers, 2 for the guards (they played a 3-guard lineup)
UK: 2 from the C, 1 from the PF, 5 from the SF (Poythress the freak show), and 1 from the PG
FGCU: 7 from the forwards, 6 from the guards (they played a 3-guard lineup)
Minn: 4 from the C, 4 from the PF, 6 from the guards (they played a 3-guard lineup)
VCU: 2 from the C, 5 from the PF, 8 from the guards (they played a 3-guard lineup)
Lou: 2 from the C, 7 from the PF, 1 from the SF, 3 from the guards
OSU: 6 from the C, 0 from the PF, 4 from the SF, 6 from the guards
As you can see, there is a LOT of guard work crashing the glass. The OSU and UK games are the ones in which we faced a substantial size difference at SF (Thompson/Ross for OSU; Poythress for UK) and in those games the SF punished us on the offensive glass. But we've allowed 28 offensive rebounds to guards and 57 offensive rebounds to forwards/centers. 28 offensive rebounds (4 per game) by guards is a pretty large number, and does suggest that it's not simply a lack of size at SF that is hurting us on the offensive glass.
I thought I'd actually look at the defensive rebounding a little closer to see where the breakdowns came from. Because just because, for example, the opposing SF got an offensive rebound doesn't necessarily mean it was our SF's fault, or his responsibility on that particular play.
This is not meant to be negative or nitpicking or anything. The team has played fabulously so far this year, and I for one am thrilled by what I've seen. One area that has not been fantastic, by the eyeball test, has been our defensive rebounding, in particular against Ohio State, and in particular in the first half. Giving up offensive rebounds can be a very discouraging thing after playing good D for most or all of a possession.
So I thought I'd take a look at the rebounding against the Buckeyes. If there's interest from you guys in this kind of thing, maybe I'd go back and do the prior games too and try to accumulate some numbers. In any event, here's what I've got from the OSU game. And again, when assigning a lapse to a particular player, I'm not trying to fingerpoint or be negative -- just assessing the plays.
To start with the numbers, we gave up 16 offensive rebounds in this game, 10 in the first half. Only 2 in the last 12 minutes of the game.
Of the 16, "responsibility" or "fault" if you will, broke down like this:
Mason: 4
Ryan: 3
Nobody: 3
Tyler: 2
Quinn: 1
Seth: 1
Rasheed: 1
Multiple guys: 1
Here's how these plays occurred, and why I assessed them the way I did. The first number is the time on the game clock when the play happened.
1922: OSU's first possession. Thompson drives baseline, shoots an airball clear over the hoop. Rasheed was totally not watching his man Smith who went right behind him from the perimeter on the weakside for an easy putback and an and-one. Responsibility: Sulaimon
1333: Smith took a long 3 from the side, hit backboard only, caromed off to the right. Ravenel got a hand on it, accidentally tipped it towards Craft, who beat Cook to it. Cook was in the lane, didn't box anyone out, but Craft was 25 or 30 feet away. Just a bad bounce. OSU did hit a 3 pointer off the play though. Responsiblity: nobody
930: Mason shaded over to help on the penetrating Scott, who had beaten Cook. Scott shot the J from the elbow and so Mason wasn't back in position to block out. But it was Thornton who had rotation responsibilities here to block out Williams, and he didn't do it. Instead, he stayed outside on Craft. The ball caromed right to Williams, who was fouled on the follow shot and hit 1 of 2. Responsibility: Thornton
858: Mason helped off on the drive by Craft. Mason contested Craft's shot, and Williams went right down the lane to grab the board. Nobody put a body on him -- Cook, Thornton, Curry. Led to an open 3, which was missed. Responsibliity: multiple guys
811: Craft drives, good position D by Cook. Craft goes up, gets it over Quinn and Mason and somehow gets the rebound right underneath the hoop. Mason never jumped for it for some reason. This led to a missed 3 by Smith. Responsibility: Mason.
807: See end of previous play. Smith missed the 3, but Ryan failed to block out Ross; no body on him at all. Ross out-athletes and outfights him for the ball. Responsibility: Kelly. Ross's follow shot was blocked by Mason, leading to a loose ball that Scott retrieved on the 3 point line.
802: See end of previous play: Scott gets this loose ball which counts as an offensive rebound. Nobody's fault - just a ball that bounced their way. Doesn't lead to points. Responsiblity: Nobody.
738: Ross makes a 1-on-1 move against Ryan, takes him to the hoop, spins and throws it up and misses but Ryan's momentum (and a lowered shoulder by Ross) takes him out of bounds beyond the baseline, leaving Ross with the easy putback of his own shot. Responsibility: Kelly.
602: Ross drove by Hairston, and Mason neither contested Ross's runner nor blocked out Williams, who got the rebound. Williams then got his follow layup sorta blocked and we fouled on the next rebounding play. Responsibility: Mason.
117: Scott drove the lane past Cook and took a runner in the lane. Again, Mason doesn't contest the shot really but instead halfway sorta moves at him, leaving him out of position for the box out, resulting in a 50-50 ball with Williams, who gets it. Williams goes up with the follow, and Mason fouls him. Made both free throws. Responsiblity: Mason.
Second half:
1408: Thompson shoots a 3 from the top. We had earlier had to switch, so Cook no longer was on Craft. Long rebound and Craft hustled and got it. Nobody's fault but Rasheed was not real quick to the long rebound. Responsibility: nobody.
1340: Curry totally failed to block out Thompson coming in from the wing to follow Craft's drive. Easy putback dunk. Responsibility: Curry.
1225: Mason helped and contested a drive by Craft, forcing a miss. But Cook didn't switch/rotate for the blockout of Williams. Luckily, he missed the follow shot. Responsibility: Cook.
114: Craft missed a 3. Mason didn't put a body on Williams, who outfought him for the rebound. Fouled on the follow - Williams hit 1 of 2. Responsiblity: Mason.
043: Thornton failed to block out Ross at all. Ross got the offensive board easily, and Thornton then fouled him. Ross hit 2. Responsibility: Thornton.
So there you have it. I know these are just raw numbers and have to be viewed in the context of the number of available rebounds there were, for how many available rebounds was each guy even in the game, and many other factors. But this info we do have now is better than nothing, I thought.
CDu, thanks for this post in the Phase I thread on the issue of defensive rebounding/opponents offensive boards:
I thought I'd actually look at the defensive rebounding a little closer to see where the breakdowns came from. Because just because, for example, the opposing SF got an offensive rebound doesn't necessarily mean it was our SF's fault, or his responsibility on that particular play.
This is not meant to be negative or nitpicking or anything. The team has played fabulously so far this year, and I for one am thrilled by what I've seen. One area that has not been fantastic, by the eyeball test, has been our defensive rebounding, in particular against Ohio State, and in particular in the first half. Giving up offensive rebounds can be a very discouraging thing after playing good D for most or all of a possession.
So I thought I'd take a look at the rebounding against the Buckeyes. If there's interest from you guys in this kind of thing, maybe I'd go back and do the prior games too and try to accumulate some numbers. In any event, here's what I've got from the OSU game. And again, when assigning a lapse to a particular player, I'm not trying to fingerpoint or be negative -- just assessing the plays.
To start with the numbers, we gave up 16 offensive rebounds in this game, 10 in the first half. Only 2 in the last 12 minutes of the game.
Of the 16, "responsibility" or "fault" if you will, broke down like this:
Mason: 4
Ryan: 3
Nobody: 3
Tyler: 2
Quinn: 1
Seth: 1
Rasheed: 1
Multiple guys: 1
Here's how these plays occurred, and why I assessed them the way I did. The first number is the time on the game clock when the play happened.
1922: OSU's first possession. Thompson drives baseline, shoots an airball clear over the hoop. Rasheed was totally not watching his man Smith who went right behind him from the perimeter on the weakside for an easy putback and an and-one. Responsibility: Sulaimon
1333: Smith took a long 3 from the side, hit backboard only, caromed off to the right. Ravenel got a hand on it, accidentally tipped it towards Craft, who beat Cook to it. Cook was in the lane, didn't box anyone out, but Craft was 25 or 30 feet away. Just a bad bounce. OSU did hit a 3 pointer off the play though. Responsiblity: nobody
930: Mason shaded over to help on the penetrating Scott, who had beaten Cook. Scott shot the J from the elbow and so Mason wasn't back in position to block out. But it was Thornton who had rotation responsibilities here to block out Williams, and he didn't do it. Instead, he stayed outside on Craft. The ball caromed right to Williams, who was fouled on the follow shot and hit 1 of 2. Responsibility: Thornton
858: Mason helped off on the drive by Craft. Mason contested Craft's shot, and Williams went right down the lane to grab the board. Nobody put a body on him -- Cook, Thornton, Curry. Led to an open 3, which was missed. Responsibliity: multiple guys
811: Craft drives, good position D by Cook. Craft goes up, gets it over Quinn and Mason and somehow gets the rebound right underneath the hoop. Mason never jumped for it for some reason. This led to a missed 3 by Smith. Responsibility: Mason.
807: See end of previous play. Smith missed the 3, but Ryan failed to block out Ross; no body on him at all. Ross out-athletes and outfights him for the ball. Responsibility: Kelly. Ross's follow shot was blocked by Mason, leading to a loose ball that Scott retrieved on the 3 point line.
802: See end of previous play: Scott gets this loose ball which counts as an offensive rebound. Nobody's fault - just a ball that bounced their way. Doesn't lead to points. Responsiblity: Nobody.
738: Ross makes a 1-on-1 move against Ryan, takes him to the hoop, spins and throws it up and misses but Ryan's momentum (and a lowered shoulder by Ross) takes him out of bounds beyond the baseline, leaving Ross with the easy putback of his own shot. Responsibility: Kelly.
602: Ross drove by Hairston, and Mason neither contested Ross's runner nor blocked out Williams, who got the rebound. Williams then got his follow layup sorta blocked and we fouled on the next rebounding play. Responsibility: Mason.
117: Scott drove the lane past Cook and took a runner in the lane. Again, Mason doesn't contest the shot really but instead halfway sorta moves at him, leaving him out of position for the box out, resulting in a 50-50 ball with Williams, who gets it. Williams goes up with the follow, and Mason fouls him. Made both free throws. Responsiblity: Mason.
Second half:
1408: Thompson shoots a 3 from the top. We had earlier had to switch, so Cook no longer was on Craft. Long rebound and Craft hustled and got it. Nobody's fault but Rasheed was not real quick to the long rebound. Responsibility: nobody.
1340: Curry totally failed to block out Thompson coming in from the wing to follow Craft's drive. Easy putback dunk. Responsibility: Curry.
1225: Mason helped and contested a drive by Craft, forcing a miss. But Cook didn't switch/rotate for the blockout of Williams. Luckily, he missed the follow shot. Responsibility: Cook.
114: Craft missed a 3. Mason didn't put a body on Williams, who outfought him for the rebound. Fouled on the follow - Williams hit 1 of 2. Responsiblity: Mason.
043: Thornton failed to block out Ross at all. Ross got the offensive board easily, and Thornton then fouled him. Ross hit 2. Responsibility: Thornton.
So there you have it. I know these are just raw numbers and have to be viewed in the context of the number of available rebounds there were, for how many available rebounds was each guy even in the game, and many other factors. But this info we do have now is better than nothing, I thought.
This is excellent analysis. Thanks so much, Tommy.
I actually don't think Duke needs to make dramatic improvements to our defensive rebounding. There are times where good rotations in Coach K's scheme will lead to guards trying to box out forwards on the defensive glass. However, I think it's interesting that the plays in which Mason was responsible for not getting to the ball are almost all because he was out of position, didn't box out, and/or didn't jump. It's true that Ryan could improve as a rebounder (his rebound rate is pretty low for a forward), I have often noticed that Mason relies on his (admittedly impressive) athleticism too often for rebounds and often fails to get good position. Luckily, he is athletic enough to make up for his lack of positioning instincts. As much as we lament having someone who can consistently rebound at the three spot, there is room for improvement for everyone on the team. Honestly, I wonder how insane Mason's numbers would be if he improves his positioning just a bit. He's already one of the best in the country.
loran16
11-30-2012, 02:25 PM
Tommy, great work. One minor complaint - it tells us nothing to say that X rebounds were the fault of player Y - what we need to see is the percentage of opportunities for rebounds that the player screwed up. So Mason was inevitably going to lead this analysis because he's the primary rebounder - he goes after the ball the most, so he's going to miss the most opportunities, but this doesn't make him inefficient. This is why I think the #s you give do indict Ryan a bit.
BTW, Tommy are you going to be doing defensive tracking again this year?
This is excellent analysis. Thanks so much, Tommy.
I actually don't think Duke needs to make dramatic improvements to our defensive rebounding. There are times where good rotations in Coach K's scheme will lead to guards trying to box out forwards on the defensive glass. However, I think it's interesting that the plays in which Mason was responsible for not getting to the ball are almost all because he was out of position, didn't box out, and/or didn't jump. It's true that Ryan could improve as a rebounder (his rebound rate is pretty low for a forward), I have often noticed that Mason relies on his (admittedly impressive) athleticism too often for rebounds and often fails to get good position. Luckily, he is athletic enough to make up for his lack of positioning instincts. As much as we lament having someone who can consistently rebound at the three spot, there is room for improvement for everyone on the team. Honestly, I wonder how insane Mason's numbers would be if he improves his positioning just a bit. He's already one of the best in the country.
Very much disagree here. In 2010 and 2011, Duke was only average at defensive rebounding, while still a top 10 defensive team. That's clearly acceptable. This year however, they're far below average. That's the problem. This isn't a small data point - this is one of the four most important factors and we are failing miserably at it. We've gotta fix that if we want to go back to being an elite defensive team. And the O can't do it alone.
Tommy, great work. One minor complaint - it tells us nothing to say that X rebounds were the fault of player Y - what we need to see is the percentage of opportunities for rebounds that the player screwed up. So Mason was inevitably going to lead this analysis because he's the primary rebounder - he goes after the ball the most, so he's going to miss the most opportunities, but this doesn't make him inefficient. This is why I think the #s you give do indict Ryan a bit.
BTW, Tommy are you going to be doing defensive tracking again this year?
Very much disagree here. In 2010 and 2011, Duke was only average at defensive rebounding, while still a top 10 defensive team. That's clearly acceptable. This year however, they're far below average. That's the problem. This isn't a small data point - this is one of the four most important factors and we are failing miserably at it. We've gotta fix that if we want to go back to being an elite defensive team. And the O can't do it alone.
I don't think I phrased my first line very well. I meant to say something along the lines of small adjustments will lead to a modest reduction in opposing team's offensive boards which in turn will lead to dramatic results. Then I wanted to use Mason as an example of how even our best rebounder can get much, much better. Poor self editing.
tommy
11-30-2012, 03:15 PM
Tommy, great work. One minor complaint - it tells us nothing to say that X rebounds were the fault of player Y - what we need to see is the percentage of opportunities for rebounds that the player screwed up. So Mason was inevitably going to lead this analysis because he's the primary rebounder - he goes after the ball the most, so he's going to miss the most opportunities, but this doesn't make him inefficient. This is why I think the #s you give do indict Ryan a bit.
BTW, Tommy are you going to be doing defensive tracking again this year?
I agree that these raw numbers don't tell the whole story. That's why I wrote what I did at the end of my post about having numbers on available rebounds and that kind of thing. But that kind of analysis would be close to impossible to do. Not only would you have to determine the number of rebounding plays that the player was on the court for, but then did he have box-out responsibilities? Was he involved in the play? Did the ball bounce anywhere near him such that he would've/could've gotten the board or would that have been unreasonable to expect? If a guy boxes out well but the ball rebounds onto the other side of the basket, and Duke gets it, and would've gotten it no regardless of his having boxed out, should the guy in question get some kind of "credit" for that? Not really. Did a guy fail to box out at all, but get lucky that the ball bounced to someone else so his failure had no impact? See what I mean? Lots of factors that would be very, very difficult to fairly and properly account for. But I'm open to suggestions!
As to your other question, I'm not going to be doing the defensive tracking again. It was very, very time consuming, and frankly, I was not satisfied that the formulas (come up with by others) really captured the essence of good and bad defense very effectively. A lot of the same type of issues that I talk about in the above paragraph vis-a-vis rebounding also applied to the defensive analysis. Defense is very, very hard to quantify. Coming up with formulas based primarily on the number of baskets and free throws your man shot and scored, as well as turnovers caused, in the context of the possessions you played, doesn't account for so much of what makes a good or bad defender. Ball pressure, stopping dribble penetration, proper hedging on a screen, help-and-recover, ball denial on the wing, lots of stuff that just wasn't getting quantified. So while I thought that using those formulas told us something about our defense, it didn't tell us enough (in my opinion) to justify the gargantuan time commitment it took to break the games down that way.
In the early part of last year, I did some play-by-play breakdowns of our defense that were instructive (kind of like the original post in this thread vis-a-vis rebounding) but I didn't get the sense that the guys on the board were so interested in that, so I stopped doing it.
The other issue I had planned on looking at this year was post/entry passing, given how important we knew Mason was going to be and what a bad passing team we were last year. But I didn't get right on it (though I still could) and it doesn't seem like that's been a problem area for us thus far, which has also limited my motivation to go to the film and analyze it. If guys were interested in that, maybe I could still do it though.
I thought I'd actually look at the defensive rebounding a little closer to see where the breakdowns came from. Because just because, for example, the opposing SF got an offensive rebound doesn't necessarily mean it was our SF's fault, or his responsibility on that particular play.
Thanks for taking the time to do this, Tommy. Good stuff. I only have a few comments to add. But I want to reiterate my thanks for you taking the time to put this together. Interesting stuff. Here are my thoughts:
811: Craft drives, good position D by Cook. Craft goes up, gets it over Quinn and Mason and somehow gets the rebound right underneath the hoop. Mason never jumped for it for some reason. This led to a missed 3 by Smith. Responsibility: Mason.
I'd add Cook to the list of responsible parties, as (assuming he was in good defensive position on the shot) Craft was his man. I think it's a bit harsh to blame Mason (even if he didn't jump) when Cook didn't seal his man. If Cook was out of position on defense, then I'd understand putting the full blame on Mason here.
802: See end of previous play: Scott gets this loose ball which counts as an offensive rebound. Nobody's fault - just a ball that bounced their way. Doesn't lead to points. Responsiblity: Nobody.
Caveat: I'm never happy with a "nobody" designation when it comes to allowing an offensive rebound. The first "nobody" you mentioned (when Ravenel tipped the ball out to Craft) is one I'll begrudgingly let slide, and this one is also borderline. Was Scott left open to get the rebound of the blocked shot? If so, that's his man's fault.
1408: Thompson shoots a 3 from the top. We had earlier had to switch, so Cook no longer was on Craft. Long rebound and Craft hustled and got it. Nobody's fault but Rasheed was not real quick to the long rebound. Responsibility: nobody.
Long rebounds are guards' responsibility. Whoever was guarding Craft on the switch (sounds like it was Sulaimon?) should get the blame. For comparison, on offense, if you're a guard on the perimeter, your first responsibility when a shot goes up is to make sure your team is protected from a runout. So the inclination is to drift back. When you're a guard on defense, your first priority when a shot is taken is to make sure the offense doesn't get the ball. Long rebounds should still go to the defense (just more likely to go to the defensive guards), and if an opposing guard got there first that's still someone's fault, in my opinion.
loran16
11-30-2012, 10:16 PM
I agree that these raw numbers don't tell the whole story. That's why I wrote what I did at the end of my post about having numbers on available rebounds and that kind of thing. But that kind of analysis would be close to impossible to do. Not only would you have to determine the number of rebounding plays that the player was on the court for, but then did he have box-out responsibilities? Was he involved in the play? Did the ball bounce anywhere near him such that he would've/could've gotten the board or would that have been unreasonable to expect? If a guy boxes out well but the ball rebounds onto the other side of the basket, and Duke gets it, and would've gotten it no regardless of his having boxed out, should the guy in question get some kind of "credit" for that? Not really. Did a guy fail to box out at all, but get lucky that the ball bounced to someone else so his failure had no impact? See what I mean? Lots of factors that would be very, very difficult to fairly and properly account for. But I'm open to suggestions!
As to your other question, I'm not going to be doing the defensive tracking again. It was very, very time consuming, and frankly, I was not satisfied that the formulas (come up with by others) really captured the essence of good and bad defense very effectively. A lot of the same type of issues that I talk about in the above paragraph vis-a-vis rebounding also applied to the defensive analysis. Defense is very, very hard to quantify. Coming up with formulas based primarily on the number of baskets and free throws your man shot and scored, as well as turnovers caused, in the context of the possessions you played, doesn't account for so much of what makes a good or bad defender. Ball pressure, stopping dribble penetration, proper hedging on a screen, help-and-recover, ball denial on the wing, lots of stuff that just wasn't getting quantified. So while I thought that using those formulas told us something about our defense, it didn't tell us enough (in my opinion) to justify the gargantuan time commitment it took to break the games down that way.
Fair enough - I'm not sure if you follow baseball, but in baseball, defensive measures are the biggest problems for those people doing statistical anaylses: some defensive metrics are like what you were doing - they involve tracking every play and it's thus hard to tell whether such metrics are accurate. Others attempt to infer from the box scores, which is just as difficult.
I happen to think the tracking was useful, if obviously not without limitations.
tommy
12-11-2012, 03:16 AM
I know there's discussion on the current KenPom thread and I think some others too -- maybe the Phase II thread if I'm not mistaken -- about our defensive rebounding. I wanted to follow up what I started in this thread after the Ohio State game and post the play-by-play if you will on this issue in the Temple game.
I am a numbers guy in a lot of respects, and I certainly respect what KenPom is doing and understand the attempts to try to calculate rebound percentages and the like. But like I said in an earlier post in this thread, as with defense, rebounding is a subject that does not lend itself easily to advanced metrics, at least in my mind. When I watch and rewatch the games and rewind the tape back and forth on each play on which the opponent gets credit for an offensive rebound, I have to say that I think the problem is being overstated. There simply weren't that many plays, at least in the Temple game this time, that we did a poor job on the defensive board. That's not to say we can't get better in this area. We can. But I'm not seeing this as some kind of glaring weakness.
Here are the Temple offensive rebounding plays.
(1st two I can't evaluate as they occurred before the 16:00 mark of the first half, when ESPN finally cut away from whatever else they were showing and got us to the Duke game, so I didn't see those plays.)
14:39 - Lee in the post against Mason. Quinn helps, and Hollis-Jefferson, who is Ryan's man, cuts down the lane. Ryan helps on Lee too, feeing H-J to get inside rebounding position and tip the rebound around. Lee hustles for the loose ball, and Seth just kinda got outfought a little bit by Lee to get it. I put responsibility for this one on Kelly, though, as he's the one who got boxed out to begin with.
14:39 - ESPN has Lee also getting credit for an offensive rebound on this one. H-J never possessed it or even controlled it. Either he or Lee should get credit for an offensive rebound for this little sequence, but not both.
14:00 - Temple on the secondary break. Randall goes to the hoop with Ryan trailing, understandably, as he had been knocked down under the hoop at the other end. Mason doesn't step in on Randall but also doesn't block out Lee. Randall's runner misses. Mason and Quinn jump for it, in good position to get it, but Lee also makes an excellent effort and gets a hand on it and bats it off the board. His second jump is quicker than Mason's, and he gets it, and scores on the follow. Responsibility: Mason, for failing to body Lee or box him and being slower on the second jump.
11:43 - O'Brien missed the 3, we have 4 guys in the lane boxing. But the long rebound goes over everyone's head, Temple gets a hand on the loose ball, and it goes out of bounds. Refs call it off Duke, but on replay it looks like they got it wrong. Have to attribute this one to "team" but was really nobody's fault, and should've been our ball in any event.
11:16 - Wyatt takes a tough turnaround from the corner with Rasheed right in his face. Tyler is boxing Randall but Randall is too strong for him and pushes Ty under the hoop, clearing space for H-J to come down the lane. H-J was Amile's man, though, and Jefferson failed to box him and lost sight of him. Quinn could've put a body on H-J too, but didn't. But primary responsibility for this one is on Amile.
8:15 - Wyatt drives into the lane and kicks to H-J, who misses from the corner. After the dish, Seth, who was in bad position in the lane after Wyatt had driven past him, didn't do anything to get in better position for the box or at least push Wyatt or anything. So Wyatt was in great position for the tip of H-J's jumper. Just missed the tip. Responsibility for the offensive board though was Seth's, in my opinion.
6:39 - Wyatt penetrates into the lane past Seth, all the way to the hoop. Mason alters his shot. Meantime, on the weakside, Ryan got caught by a semi-intentional back screen and his man Pepper went to crash. Problem was that Rasheed didn't call the screen on Ryan until it was too late for either himself or Ryan to try to stay with Pepper and block him out. Pepper got the easy board. Responsibility: Sulaimon.
1:36 - Pepper missed a transition 3. H-J came tearing down the lane, hustled after the rebound and saved it as it was going over the end line and threw it back out towards the 3 point line. It was an out of control "pass" though, and it ended up going off of DeCosey and out of bounds. Somehow ESPN had this down as not one but two offensive rebounds. One for H-J and one for DeCosey. Should have been zero, not two, credited on this play.
0:39 - ESPN's play by play had an Offensive Rebound here credited to Temple's team, but in my review there was no such play anywhere near this point in the game.
Second half:
11:38 - Out of bounds play for Temple. Pepper's corner 3 is off. Amile failed to put a body on Randall, who pushed him out of the way for the board and kick out. Responsibility: Jefferson.
11:26 - DiLeo takes a runner in the lane, Alex jumps to try to block it. Somehow DiLeo's momentum carried him to the other side of the basket and inside of Mason, and Amile was out of position too. DiLeo gets his own board and putback. I have to put this one on Mason, as there's no way a guy this much shorter should get an offensive rebound in your vicinity no matter what.
9:03 - O'Brien shoots a 3 from the top. Sulaimon is covering Randall. Randall floats out away from the basket instead of crashing, so Rasheed is in great position for the board, but he, Ryan, and a Temple guy all get their hands on the board and it bounces out and Randall then busts back in and beats Mason to it in the lane. I think this one goes down as "team" responsibility.
8:49 - continuation of above play. Randall jacks up an immediate 3 -- bad shot. Mason totally fails to block out O'Brien, who comes right down Broadway for the putback and the and-one.
2:00 - Murphy is beaten base and Ryan blocks the layup out of bounds. Why does this count as an offensive rebound for Temple? I don't think it does, or should.
0:47 - Josh helps on a drive, leaving his man Dingle for a follow tip. Really should've been Rasheed's help on the driver so Josh wouldn't have had to leave his man. Murphy on the weakside had no chance. Responsibility: Hairston and Sulaimon.
So, adding it up, I count a total of 11 offensive rebounds that we gave up (again, I missed the first two of the game thanks to ESPN) and responsibility for the 11 went like this:
Mason: 3
Amile: 2
Team: 2
Rasheed: 1 1/2
Ryan: 1
Seth: 1
Josh: 1/2
Guys, I don't think that's so bad. It's not so many, really, especially given the style of defense we play. And that defense prevented most of these offensive rebounds that we did give up from being converted into baskets. If you were to look at it from a per-minutes perspective, the only one who maybe warrants a little concern is Amile, but his lapses are due to inexperience and lack of strength, both of which are to be expected.
timmy c
12-11-2012, 08:28 AM
Tommy,
Great analysis!
I went back and looked at the first two offensive rebounds.
15:58 – Randall misses a midrange jumper off the back of the rim that bounces high into the air. Hairston has position, but Lee jumps higher while knocking the ball back to an area where Randall recovers. Randall gets credit for the Rebound. Responsibility: Hairston , Suilamon for not boxing out the shooter.
15:25 – Seth gets beat off the dribble. Kelly rotates to stop the ball. Cook rotates to the middle but never puts a body on Kelly's man, H-J.
H-J, dunks the rebound. Responsibility: Cook
Mason: 3
Amile: 2
Team: 2
Rasheed: 2
Ryan: 1
Seth: 1
Josh: 1
Cook: 1
14:39 - ESPN has Lee also getting credit for an offensive rebound on this one. H-J never possessed it or even controlled it. Either he or Lee should get credit for an offensive rebound for this little sequence, but not both
ESPN credited Hollis-Jefferson's tip as a shot attempt. Therefore, he got a rebound and a missed shot. Lee's rebound was credited on Hollis-Jefferson's shot attempt. If you want to take away Hollis-Jefferson's rebound, then you have to also take away his shot attempt.
1:36 - Pepper missed a transition 3. H-J came tearing down the lane, hustled after the rebound and saved it as it was going over the end line and threw it back out towards the 3 point line. It was an out of control "pass" though, and it ended up going off of DeCosey and out of bounds. Somehow ESPN had this down as not one but two offensive rebounds. One for H-J and one for DeCosey. Should have been zero, not two, credited on this play.
First of all, it should have been one, not zero, rebounds. Second, I think this is an example of the play-by-play not being consistent with the box score. Notice that ESPN's box score does not credit DeCosey with any offensive rebounds. So clearly the box score got it right, and the play-by-play did not. Still, if Hollis-Jefferson got the ball and then threw it off DeCosey and out of bounds, that should be an offensive rebound for Hollis-Jefferson and a turnover. So maybe a "team" fault?
0:39 - ESPN's play by play had an Offensive Rebound here credited to Temple's team, but in my review there was no such play anywhere near this point in the game.
O'Brien missed a 3 and the ball went out of bounds. When this happens, the standard rule is that the team who gains possession gets credited with a team rebound. Since the ball apparently went off us, Temple got a team rebound. Perhaps blame this on "team?"
So, adding it up, I count a total of 11 offensive rebounds that we gave up (again, I missed the first two of the game thanks to ESPN) and responsibility for the 11 went like this:
Mason: 3
Amile: 2
Team: 2
Rasheed: 1 1/2
Ryan: 1
Seth: 1
Josh: 1/2
Guys, I don't think that's so bad. It's not so many, really, especially given the style of defense we play. And that defense prevented most of these offensive rebounds that we did give up from being converted into baskets. If you were to look at it from a per-minutes perspective, the only one who maybe warrants a little concern is Amile, but his lapses are due to inexperience and lack of strength, both of which are to be expected.
Adding back the one that you didn't credit to Hollis-Jefferson on the fast break, the one you didn't credit to Hollis-Jefferson on the tip, and the team rebound at 0:39 of the 1st, that's 14 plus the 2 in the first 5 minutes. That's 16 of 47(?), or 34%. If you don't want to count the Hollis-Jefferson missed tip as a rebound, then it is 15 of 46 (33%). That's slightly better than the 35% we were allowing in previous games, but it is still not very good. As someone noted earlier, the 25th best defensive rebounding team allows only 26.6% of missed shots to be offensive rebounds. So I'd say we still fell on the below-average side of things overall.
tommy
12-11-2012, 01:25 PM
ESPN credited Hollis-Jefferson's tip as a shot attempt. Therefore, he got a rebound and a missed shot. Lee's rebound was credited on Hollis-Jefferson's shot attempt. If you want to take away Hollis-Jefferson's rebound, then you have to also take away his shot attempt.
First of all, it should have been one, not zero, rebounds. Second, I think this is an example of the play-by-play not being consistent with the box score. Notice that ESPN's box score does not credit DeCosey with any offensive rebounds. So clearly the box score got it right, and the play-by-play did not. Still, if Hollis-Jefferson got the ball and then threw it off DeCosey and out of bounds, that should be an offensive rebound for Hollis-Jefferson and a turnover. So maybe a "team" fault?
O'Brien missed a 3 and the ball went out of bounds. When this happens, the standard rule is that the team who gains possession gets credited with a team rebound. Since the ball apparently went off us, Temple got a team rebound. Perhaps blame this on "team?"
Adding back the one that you didn't credit to Hollis-Jefferson on the fast break, the one you didn't credit to Hollis-Jefferson on the tip, and the team rebound at 0:39 of the 1st, that's 14 plus the 2 in the first 5 minutes. That's 16 of 47(?), or 34%. If you don't want to count the Hollis-Jefferson missed tip as a rebound, then it is 15 of 46 (33%). That's slightly better than the 35% we were allowing in previous games, but it is still not very good. As someone noted earlier, the 25th best defensive rebounding team allows only 26.6% of missed shots to be offensive rebounds. So I'd say we still fell on the below-average side of things overall.
What I'm trying to do is to look at the forest rather than get caught up in the trees. By that I mean: what is really going on with our rebounding when you actually sit down and watch these plays? Is our actual rebounding being accurately represented by the percentages quoted by KenPom and others? I don't really think so.
For instance, to take the last one first, O'Brien misses a 3-ball and the ball goes out of bounds and is awarded to Temple. Somebody somewhere has decided that they deserve credit for an offensive rebound on a play like that. That counts "against" Duke's defensive rebounding percentage. But that's not the kind of play that is, or should be, causing any kind of concern as to Duke's ability to defensively rebound the ball. Without having it to review this second, if I recall it correctly -- and I may not be -- I think it was a long clanger that just bounced a certain way and both teams went after it -- on the floor -- and it went off Duke. I don't think of that as a poor rebounding play on Duke's part. Maybe we got outhustled on the floor, but to me, that's a different issue.
By the same token, H-J's play at 1:36 was not a play I think of as a "rebounding" play. The ball was careening out of bounds and he jumped beyond the baseline and made a desperation save and just threw it back out into play. Hardly the type of play that indicates a problem in our rebounding. It was a save and an attempted pass by H-J that ended up going out of bounds off his own man. Just because a player touches a ball like H-J did here, and even controlled it for a moment, that doesn't necessarily mean it was a true offensive rebounding play or a play that indicates anything about our defensive rebounding. I mean, you could say, I suppose, that someone should have blocked H-J out as he tore down from halfcourt trailing the break, but as you know that's almost impossible to do when a guy has a head of steam and the defenders are busy trying to stop the initial break anyway.
The play at 14:39, H-J got a finger on an available offensive board, so sure, give him credit. And I did. But his tipping the ball and creating a loose ball out of it on the floor does not, in my mind, create a new and separate offensive rebounding opportunity. At that point it's a loose ball. Lee's getting that loose ball was, again, not an offensive rebounding play in the way that I think most of us are discussing or even concerned about Duke's defensive rebounding. Again, maybe a hustle and get down on the floor issue, but that's different. It's not rebounding.
So when I look at these and a few others, and I look at how many times were there actual, true offensive rebounds as most folks probably think of when they think of an offensive rebounding opportunity, where the defender fails to block out or gets pushed out of the way or outfought for a ball off the glass, that kind of thing, there really weren't that many. Again, I didn't/couldn't evaluate the first two on ESPN's play by play (and you're right, their play by play often doesn't match their box score) but as for the final 36 minutes of the game, I only count 10 or 11, far less than the 14 or 15 that ESPN had for those minutes. That's a statistically significant difference. Not sure where KenPom is getting his numbers that he plugs into his percentages, but all I can say is that the number of plays, at least in this game, that got counted as offensive rebounds for Temple was significantly greater than the number of plays that evidenced poor defensive rebounding by Duke.
What I'm trying to do is to look at the forest rather than get caught up in the trees. By that I mean: what is really going on with our rebounding when you actually sit down and watch these plays? Is our actual rebounding being accurately represented by the percentages quoted by KenPom and others? I don't really think so.
And I think Pomeroy's estimates are accurately assessing the situation, because the plays you are complaining about are fairly standard calls that happen in all of college basketball. So other teams suffer the same "misfortunes" that Duke has suffered. See below.
For instance, to take the last one first, O'Brien misses a 3-ball and the ball goes out of bounds and is awarded to Temple. Somebody somewhere has decided that they deserve credit for an offensive rebound on a play like that. That counts "against" Duke's defensive rebounding percentage. But that's not the kind of play that is, or should be, causing any kind of concern as to Duke's ability to defensively rebound the ball. Without having it to review this second, if I recall it correctly -- and I may not be -- I think it was a long clanger that just bounced a certain way and both teams went after it -- on the floor -- and it went off Duke. I don't think of that as a poor rebounding play on Duke's part. Maybe we got outhustled on the floor, but to me, that's a different issue.
This example isn't a case of "somebody somewhere deciding" that Temple deserved a rebound. That's simply how the stat is applied universally. When a shot is missed, someone gets credit for a rebound. If no individual warrants credit (i.e., the ball went out of bounds) then the team is credited with the rebound. This is not some anomalous situation where Duke is getting hosed. It happens quite regularly.
By the same token, H-J's play at 1:36 was not a play I think of as a "rebounding" play. The ball was careening out of bounds and he jumped beyond the baseline and made a desperation save and just threw it back out into play. Hardly the type of play that indicates a problem in our rebounding. It was a save and an attempted pass by H-J that ended up going out of bounds off his own man. Just because a player touches a ball like H-J did here, and even controlled it for a moment, that doesn't necessarily mean it was a true offensive rebounding play or a play that indicates anything about our defensive rebounding. I mean, you could say, I suppose, that someone should have blocked H-J out as he tore down from halfcourt trailing the break, but as you know that's almost impossible to do when a guy has a head of steam and the defenders are busy trying to stop the initial break anyway.
Again - it doesn't matter what you think. That's simply the way the stats are applied universally. Happens fairly frequently.
The play at 14:39, H-J got a finger on an available offensive board, so sure, give him credit. And I did. But his tipping the ball and creating a loose ball out of it on the floor does not, in my mind, create a new and separate offensive rebounding opportunity. At that point it's a loose ball. Lee's getting that loose ball was, again, not an offensive rebounding play in the way that I think most of us are discussing or even concerned about Duke's defensive rebounding. Again, maybe a hustle and get down on the floor issue, but that's different. It's not rebounding.
And as I said, if you don't feel that it deserved a rebound, then take away both the rebound and Hollis-Jefferson's missed shot. This is the only case on your list where I think you might have a legitimate argument (and only because I haven't gone back to see the game). The scorekeepers felt the tip was worthy of being called a shot attempt. If it was, then it should count as a rebound (and missed shot). If it wasn't, then it should have been neither a missed shot nor a rebound (and we'd be at 15 of 46 instead of 16 of 47).
So when I look at these and a few others, and I look at how many times were there actual, true offensive rebounds as most folks probably think of when they think of an offensive rebounding opportunity, where the defender fails to block out or gets pushed out of the way or outfought for a ball off the glass, that kind of thing, there really weren't that many. Again, I didn't/couldn't evaluate the first two on ESPN's play by play (and you're right, their play by play often doesn't match their box score) but as for the final 36 minutes of the game, I only count 10 or 11, far less than the 14 or 15 that ESPN had for those minutes. That's a statistically significant difference. Not sure where KenPom is getting his numbers that he plugs into his percentages, but all I can say is that the number of plays, at least in this game, that got counted as offensive rebounds for Temple was significantly greater than the number of plays that evidenced poor defensive rebounding by Duke.
Well, I'm not sure that the bolded statements here are correct. Even ignoring the fact that you're arbitrarily changing the rules for rebounding for your analysis, I don't think there's a large enough sample to confidently say that 11 (instead of 14) is a statistically significant difference. But regardless of that, you're ignoring that the few rebounds for which you've discounted happen to other teams too, and that the stats were appropriately applied for the given instances. If you want to do a game-by-game analysis of ALL teams' rebounding games and adjust every team's offensive rebounds allowed down according to your criteria, feel free. But I suspect you'll find that these sort of rebounds all even out in the wash. By taking a microscope to Duke's rebound plays and combing out the ones you disagree with for only Duke, you're biasing things in favor of Duke.
timmy c
12-11-2012, 02:39 PM
Not sure where KenPom is getting his numbers that he plugs into his percentages, but all I can say is that the number of plays, at least in this game, that got counted as offensive rebounds for Temple was significantly greater than the number of plays that evidenced poor defensive rebounding by Duke.
Directly from KenPom...
OR% = OR / (OR + DRopp)
Keep in mind that rebounding percentage is computed from box score data which does not contain team rebounds. Therefore, the figures shown here may differ slightly from calculations made on the rebounding totals provided by a school
For instance, to take the last one first, O'Brien misses a 3-ball and the ball goes out of bounds and is awarded to Temple. Somebody somewhere has decided that they deserve credit for an offensive rebound on a play like that. That counts "against" Duke's defensive rebounding percentage. But that's not the kind of play that is, or should be, causing any kind of concern as to Duke's ability to defensively rebound the ball. Without having it to review this second, if I recall it correctly -- and I may not be -- I think it was a long clanger that just bounced a certain way and both teams went after it -- on the floor -- and it went off Duke. I don't think of that as a poor rebounding play on Duke's part. Maybe we got outhustled on the floor, but to me, that's a different issue.
Since this is a team rebound, KenPom would not count this in his rebound statistics. However, this clearly puts Duke at a disadvantage since the opposition gets another chance at scoring.
What I'm trying to do is to look at the forest rather than get caught up in the trees. By that I mean: what is really going on with our rebounding when you actually sit down and watch these plays? Is our actual rebounding being accurately represented by the percentages quoted by KenPom and others? I don't really think so.
KenPom’s data is not designed to measure teams’ in order to give a ranking, like RPI or the AP polls. Nor is he attempting to spend time watching every D-1 college basketball game in order to determine which rebounds are more valuable than others. Instead, he is trying to look at all the measurable data available in order to figure out which data is important for predicting future success. A realize this might be splitting hairs for some of you, but if you are attempting to rank Duke with his data, you are missing the point.
With that knowledge, it should be noted that anytime a shot misses and the team shooting the ball maintains possession, an advantage is gained for that team. If a team becomes particularly successful at this tactic, it will contribute to future wins.
This creates a rock/paper/scissors situation – team A beats team B 8/10 times; team B beats team C 8/10 times; and team C beats team A 8/10 times. If Duke is team A it would want to play against team B instead of team C. Teams that are good at offensive rebounds could be more like team C. In March, if Duke plays against a team that doesn't turn the ball over and gets a fair share of their misses, this could be enough to swing the statistical probabilities in their direction.
supbros
12-11-2012, 03:02 PM
This creates a rock/paper/scissors situation – team A beats team B 8/10 times; team B beats team C 8/10 times; and team C beats team A 8/10 times. If Duke is team A it would want to play against team B instead of team C. Teams that are good at offensive rebounds could be more like team C. In March, if Duke plays against a team that doesn't turn the ball over and gets a fair share of their misses, this could be enough to swing the statistical probabilities in their direction.
Just to quickly touch on this: Minnesota is the 10th best team according to KenPom and rank 13th and 16th in the polls. They are rebounding 47.4% (!!!) of their own misses against a reasonably difficult schedule, easily the best among all ranked teams.
If there was ever a team geared to give Duke fits, it's them. Yet we still crushed them (albeit with a little bit of help from white hot 3 pt shooting).
Right now Duke is a neutral court fave vs anybody other than Indiana (who unfortunately happens to be pretty strong on the glass)
Kedsy
12-11-2012, 03:05 PM
This creates a rock/paper/scissors situation – team A beats team B 8/10 times; team B beats team C 8/10 times; and team C beats team A 8/10 times. If Duke is team A it would want to play against team B instead of team C. Teams that are good at offensive rebounds could be more like team C. In March, if Duke plays against a team that doesn't turn the ball over and gets a fair share of their misses, this could be enough to swing the statistical probabilities in their direction.
Except Duke has played several teams that are "like team C" and has beaten them. Ultimately, you have to look at all four factors. Allowing a bunch of offensive rebounds is a weakness that can be overcome if you're really good at the other three factors.
tommy
12-12-2012, 12:58 AM
And I think Pomeroy's estimates are accurately assessing the situation, because the plays you are complaining about are fairly standard calls that happen in all of college basketball. So other teams suffer the same "misfortunes" that Duke has suffered. See below.
This example isn't a case of "somebody somewhere deciding" that Temple deserved a rebound. That's simply how the stat is applied universally. When a shot is missed, someone gets credit for a rebound. If no individual warrants credit (i.e., the ball went out of bounds) then the team is credited with the rebound. This is not some anomalous situation where Duke is getting hosed. It happens quite regularly.
Again - it doesn't matter what you think. That's simply the way the stats are applied universally. Happens fairly frequently.
And as I said, if you don't feel that it deserved a rebound, then take away both the rebound and Hollis-Jefferson's missed shot. This is the only case on your list where I think you might have a legitimate argument (and only because I haven't gone back to see the game). The scorekeepers felt the tip was worthy of being called a shot attempt. If it was, then it should count as a rebound (and missed shot). If it wasn't, then it should have been neither a missed shot nor a rebound (and we'd be at 15 of 46 instead of 16 of 47).
Well, I'm not sure that the bolded statements here are correct. Even ignoring the fact that you're arbitrarily changing the rules for rebounding for your analysis, I don't think there's a large enough sample to confidently say that 11 (instead of 14) is a statistically significant difference. But regardless of that, you're ignoring that the few rebounds for which you've discounted happen to other teams too, and that the stats were appropriately applied for the given instances. If you want to do a game-by-game analysis of ALL teams' rebounding games and adjust every team's offensive rebounds allowed down according to your criteria, feel free. But I suspect you'll find that these sort of rebounds all even out in the wash. By taking a microscope to Duke's rebound plays and combing out the ones you disagree with for only Duke, you're biasing things in favor of Duke.
You're missing the point, maybe because all you're doing, because it's all KenPom is doing, is taking the numbers out of a box score and plugging them into simple formulas and saying "Aha! Rebounding problem for Team X" Or "Team Y is a great rebounding team -- see these numbers!"
Those numbers, in my opinion -- and you obviously don't have to share it -- do not do a very good job of capturing the presence or absence of the type of rebounding problem that folks on this board seem to be concerned about vis-a-vis Duke. For two reasons. One, as I've stated in an earlier post or two, there are so many variables that go into rebounding that are simply not easily quantifiable. I've given examples upthread, and don't need to do so again. Second, the types of plays that I've mentioned in the posts regarding the Temple game that are being counted as offensive rebounds simply are not what I consider to be true offensive rebounds, and even if they technically meet some 'standard' definition out there, I don't care. They still aren't the types of plays that are indicative of a defensive rebounding 'problem' on Duke's part, or whoever the team would be who is getting these types of plays counted 'against' their defensive rebounding numbers. And because of that reality, a reality not recognized by the numbers, the numbers are not telling a particularly complete story here.
You say it doesn't matter what I think, and that's fine. But I'm entitled to my opinion, based on a very close review of every single one of these plays, that the box score-type numbers are inflating the actual numbers of what most people are thinking of when they think of what is an offensive rebound and what types of plays evidence a failure of defensive rebounding. And there are four or five examples of this just in the Temple game alone. And it is statistically significant. If instead of 15 OR's, Temple only were to have been credited with 11, that's about a 27% reduction.
If you want to hold the position that Duke has a defensive rebounding problem because opponents are throwing themselves over the endline to save loose balls and chucking them back towards halfcourt and the opponent is getting those loose balls, or because our 6'2" guard gets beaten to a loose ball on the floor after a tip play underneath, or because after our big man blocks the opponent's shot out of bounds and the opponent somehow gets credit for a "team" offensive rebound -- if you want to be worried about our defensive rebounding because of plays like that and the ridiculous way they're being counted statistically, hey, knock yourself out. I'm confident that the true scope of the "problem" -- to the extent it exists at all -- is far less than what these box score numbers would have us think.
You're missing the point, maybe because all you're doing, because it's all KenPom is doing, is taking the numbers out of a box score and plugging them into simple formulas and saying "Aha! Rebounding problem for Team X" Or "Team Y is a great rebounding team -- see these numbers!"
Those numbers, in my opinion -- and you obviously don't have to share it -- do not do a very good job of capturing the presence or absence of the type of rebounding problem that folks on this board seem to be concerned about vis-a-vis Duke. For two reasons. One, as I've stated in an earlier post or two, there are so many variables that go into rebounding that are simply not easily quantifiable. I've given examples upthread, and don't need to do so again. Second, the types of plays that I've mentioned in the posts regarding the Temple game that are being counted as offensive rebounds simply are not what I consider to be true offensive rebounds, and even if they technically meet some 'standard' definition out there, I don't care. They still aren't the types of plays that are indicative of a defensive rebounding 'problem' on Duke's part, or whoever the team would be who is getting these types of plays counted 'against' their defensive rebounding numbers. And because of that reality, a reality not recognized by the numbers, the numbers are not telling a particularly complete story here.
You say it doesn't matter what I think, and that's fine. But I'm entitled to my opinion, based on a very close review of every single one of these plays, that the box score-type numbers are inflating the actual numbers of what most people are thinking of when they think of what is an offensive rebound and what types of plays evidence a failure of defensive rebounding. And there are four or five examples of this just in the Temple game alone. And it is statistically significant. If instead of 15 OR's, Temple only were to have been credited with 11, that's about a 27% reduction.
If you want to hold the position that Duke has a defensive rebounding problem because opponents are throwing themselves over the endline to save loose balls and chucking them back towards halfcourt and the opponent is getting those loose balls, or because our 6'2" guard gets beaten to a loose ball on the floor after a tip play underneath, or because after our big man blocks the opponent's shot out of bounds and the opponent somehow gets credit for a "team" offensive rebound -- if you want to be worried about our defensive rebounding because of plays like that and the ridiculous way they're being counted statistically, hey, knock yourself out. I'm confident that the true scope of the "problem" -- to the extent it exists at all -- is far less than what these box score numbers would have us think.
Both sides of this argument have merit, but folks are really arguing different points.
Our rebounding stats, with all the noise mixed in, are used when comparing with other schools, who also have noise mixed in. Presumably, being the purpose of statistics, the noise mostly evens out so that teams can be roughly compared by numbers. Bookies like stats.
I feel certain the coaching staff, though, doesn't look at the stats but rather looks to see if a player had the right defensive position and made the proper choice under the specific circumstance. Coaches like film.
Both approaches have validity. Coach isn't teaching from stats, and no one is going to look, play by play, at 350ish schools' rebounding.
-jk
You're missing the point, maybe because all you're doing, because it's all KenPom is doing, is taking the numbers out of a box score and plugging them into simple formulas and saying "Aha! Rebounding problem for Team X" Or "Team Y is a great rebounding team -- see these numbers!"
No, I'm not missing the point. I completely understand what you're trying to do. I'm just trying to illustrate why your attempt at doing so isn't correct. The "problem" with Duke's rebounding is relative to other team's rebounding percentages. So subtracting the rebounds that you don't think are Duke's fault without doing the same for other teams to suggest that Duke doesn't have a rebounding problem isn't fair. Subtract away the similar rebounds from those other teams and you're left with the same relative rebounding problem for Duke, even if the absolute numbers are smaller. In fact, it might make the relative problem worse (hold the absolute difference the same and subtract from both totals and the relative difference is greater).
Those numbers, in my opinion -- and you obviously don't have to share it -- do not do a very good job of capturing the presence or absence of the type of rebounding problem that folks on this board seem to be concerned about vis-a-vis Duke.
Actually, I agree with you that the numbers may not exactly reflect the "fault" rebounds. My point of disagreement is that (1) you're isolating the rebounds for Duke and ignoring the same for other teams and (2) doing it wrong (see below) in an attempt to suggest that Duke doesn't really have a rebounding problem. That's not sound analysis, even if your review of the Duke tape is sound (which it is).
In other words, I'm saying that, if you do the same analysis for other teams, you'll subtract away a similar number of rebounds. And that will leave you with the same conclusion as Pomeroy - that we have a rebounding deficiency relative to other teams.
You say it doesn't matter what I think, and that's fine. But I'm entitled to my opinion, based on a very close review of every single one of these plays, that the box score-type numbers are inflating the actual numbers of what most people are thinking of when they think of what is an offensive rebound and what types of plays evidence a failure of defensive rebounding. And there are four or five examples of this just in the Temple game alone. And it is statistically significant. If instead of 15 OR's, Temple only were to have been credited with 11, that's about a 27% reduction.
You are most certainly entitled to your opinion. That doesn't change the fact that your conclusions are based on flawed analysis (see further below).
Also, I'm intrigued by how your numbers have grown from a difference of 2-3 (which also wasn't correct) to 4-5 (which definitely isn't correct) examples as this discussion has gone on. There are 14 rebounds listed in the box score (remember - the box score doesn't include the "team" rebounds, and the play-by-play isn't an exact match with the box score anyway). In your analysis, you attributed fault to Duke on 10 or 11, and that excluded the first 2 rebound plays (which someone else attributed to Duke). Subtract those 2 rebounds from the 14, and you get 12 (versus your 10 or 11). That's a difference of 1 or 2. Not 4 or 5.
Additionally, I still don't believe that a difference of 4 or 5 would be statistically significant, given the small sample size. It may seem significant, but I'm not sure that, statistically speaking, it is. But that's kind of irrelevant since the difference isn't 4 or 5, but 1 or 2 (which most certainly isn't significant).
If you want to hold the position that Duke has a defensive rebounding problem because opponents are throwing themselves over the endline to save loose balls and chucking them back towards halfcourt and the opponent is getting those loose balls, or because our 6'2" guard gets beaten to a loose ball on the floor after a tip play underneath, or because after our big man blocks the opponent's shot out of bounds and the opponent somehow gets credit for a "team" offensive rebound -- if you want to be worried about our defensive rebounding because of plays like that and the ridiculous way they're being counted statistically, hey, knock yourself out. I'm confident that the true scope of the "problem" -- to the extent it exists at all -- is far less than what these box score numbers would have us think.
And here you further illustrate that you don't get what I'm saying. I don't hold the position that Duke has a rebounding problem because of a few random plays or the rules that you disagree with. I hold the position because Duke has a relative rebounding problem compared with other teams. If you take out the rebounds that you disagree with from the Duke games AND DO THE SAME FOR ALL OTHER TEAMS TOO you'll find that Duke STILL has a relative rebounding problem, because you'll eliminate those same rebounds that happen to other teams (which will leave roughly the same margin as before).
And that's even assuming that your analysis differs mathematically from Pomeroy's. In this thread alone, we've noted multiple ways that you have not correctly done this:
1. using the play-by-pay stats as gospel and not checking them against the box score
2. failing to realize that Pomeroy doesn't include team rebounds in his analysis (since he looks only at box scores, which don't present the team rebounds)
3. counting the two offensive rebounds that you didn't see in the first 5 minutes in your discrepancy difference, rather than subtracting those rebounds from the total (you should be comparing against 12, not 15)
4. adding rebounds to the comparator list (the box score has 14 rebounds, not 15)
Again, take away those errors, and you're left with 10 or 11 rebounds in your count and 12 rebounds in the box score count. That's a difference of 1 or 2, and that is most certainly not statistically significant. And that's ignoring the fact that you are only doing this for Duke and trying to say that we don't have a rebounding problem relative to other teams.
Your analysis is well-intentioned and I certainly commend the effort. But there are multiple mistakes you've made that render your conclusion invalid.
Both sides of this argument have merit, but folks are really arguing different points.
Our rebounding stats, with all the noise mixed in, are used when comparing with other schools, who also have noise mixed in. Presumably, being the purpose of statistics, the noise mostly evens out so that teams can be roughly compared by numbers. Bookies like stats.
I feel certain the coaching staff, though, doesn't look at the stats but rather looks to see if a player had the right defensive position and made the proper choice under the specific circumstance. Coaches like film.
Both approaches have validity. Coach isn't teaching from stats, and no one is going to look, play by play, at 350ish schools' rebounding.
-jk
Tommy's point that the box score stats overstate the absolute rebounding numbers for teams is absolutely correct. Where his argument doesn't have merit is when he then takes that completely correct statement and suggests that the box scores are overstating Duke's rebounding problem (for the reasons I laid out in my previous post). Duke's rebounding problem is a relative problem, and the box score scaling issue doesn't fix that relative problem.
And please note that I don't actually suggest that tommy do the same for all other teams. I merely mentioned that to show that the jump from his first statement (box scores may inflate absolute rebound numbers) to his second statement (Pomeroy's analysis of box scores inflates Duke's rebounding problem) is invalid.
NBA statisticians have recently started trying to assign different values for various types of assists. A pass to a cutter who gets an easy dunk or layup has more value than a pass to a half-way open shooter who takes a contested long two point shot. Passes leading to open threes from the corner (the highest value shot in the NBA due to the relatively close distance compared to three point shots elsewhere on the court) are more valuable than passes leading to threes from the top of the key. The idea is that the creator that manufactures more layups and corner threes from his passes has a higher value than someone who only generates lower value jump shots and longer threes. There are certainly some flaws in this idea, but it still is an interesting way to differentiate between the different types of assists.
An interesting stat to follow for Duke this season might be the percentage of offensive rebounds allowed that lead to scores by the defense. Possibly a better stat would be points per offensive rebound. If Duke is allowing offensive boards, but only doing so because our rotations are taking people out rebounding position and the defense recovers in time to force another missed shot or at least a contested shot, those are offensive boards that have far less value than Orebs leading directly to easy put backs or wide open threes on kick-outs. By measuring the points allowed per offensive rebound, we can get a basic picture of how much those rebounds actually hurt the team.
Using the 2nd chance points found in the GoDuke.com box scores, through nine games Duke has given up 123 points from 127 offensive rebounds or 0.97 points per offensive rebound. I have no point of comparison for this stat (perhaps I will have the time and energy to compute this stat for past Duke teams to use as a measuring stick), but this would seem to indicate to me that the majority of the offensive rebounds allowed have been of the lower-value variety. It also indicates that our defense really has been excellent in the half-court, as even the extra opportunities our opponents have gotten do not often lead to easy shots. The key thing here is that Duke's defense must continue to force opponents into bad (or at least sub-par) shooting nights, which, so far, we've done against everyone we've faced.
So, the notion that an offensive rebound comes from a good thing, a missed shot, certainly has some merit provided Duke can continue to play outstanding defense and give up relatively low value offensive rebounds. At the same time, our defensive numbers really would improve substantially if we started hauling in a few more defensive boards per game. If the current rate of points allowed per offensive board holds, we can value an offensive rebound by our opponent at 1 point each (rounding up) on average. In tight games, simply grabbing one or two extra defensive boards could prove VERY valuable.
tommy
12-13-2012, 12:53 AM
No, I'm not missing the point. I completely understand what you're trying to do. I'm just trying to illustrate why your attempt at doing so isn't correct. The "problem" with Duke's rebounding is relative to other team's rebounding percentages. So subtracting the rebounds that you don't think are Duke's fault without doing the same for other teams to suggest that Duke doesn't have a rebounding problem isn't fair. Subtract away the similar rebounds from those other teams and you're left with the same relative rebounding problem for Duke, even if the absolute numbers are smaller. In fact, it might make the relative problem worse (hold the absolute difference the same and subtract from both totals and the relative difference is greater).
Actually, I agree with you that the numbers may not exactly reflect the "fault" rebounds. My point of disagreement is that (1) you're isolating the rebounds for Duke and ignoring the same for other teams and (2) doing it wrong (see below) in an attempt to suggest that Duke doesn't really have a rebounding problem. That's not sound analysis, even if your review of the Duke tape is sound (which it is).
In other words, I'm saying that, if you do the same analysis for other teams, you'll subtract away a similar number of rebounds. And that will leave you with the same conclusion as Pomeroy - that we have a rebounding deficiency relative to other teams.
You are most certainly entitled to your opinion. That doesn't change the fact that your conclusions are based on flawed analysis (see further below).
Also, I'm intrigued by how your numbers have grown from a difference of 2-3 (which also wasn't correct) to 4-5 (which definitely isn't correct) examples as this discussion has gone on. There are 14 rebounds listed in the box score (remember - the box score doesn't include the "team" rebounds, and the play-by-play isn't an exact match with the box score anyway). In your analysis, you attributed fault to Duke on 10 or 11, and that excluded the first 2 rebound plays (which someone else attributed to Duke). Subtract those 2 rebounds from the 14, and you get 12 (versus your 10 or 11). That's a difference of 1 or 2. Not 4 or 5.
Additionally, I still don't believe that a difference of 4 or 5 would be statistically significant, given the small sample size. It may seem significant, but I'm not sure that, statistically speaking, it is. But that's kind of irrelevant since the difference isn't 4 or 5, but 1 or 2 (which most certainly isn't significant).
And here you further illustrate that you don't get what I'm saying. I don't hold the position that Duke has a rebounding problem because of a few random plays or the rules that you disagree with. I hold the position because Duke has a relative rebounding problem compared with other teams. If you take out the rebounds that you disagree with from the Duke games AND DO THE SAME FOR ALL OTHER TEAMS TOO you'll find that Duke STILL has a relative rebounding problem, because you'll eliminate those same rebounds that happen to other teams (which will leave roughly the same margin as before).
And that's even assuming that your analysis differs mathematically from Pomeroy's. In this thread alone, we've noted multiple ways that you have not correctly done this:
1. using the play-by-pay stats as gospel and not checking them against the box score
2. failing to realize that Pomeroy doesn't include team rebounds in his analysis (since he looks only at box scores, which don't present the team rebounds)
3. counting the two offensive rebounds that you didn't see in the first 5 minutes in your discrepancy difference, rather than subtracting those rebounds from the total (you should be comparing against 12, not 15)
4. adding rebounds to the comparator list (the box score has 14 rebounds, not 15)
Again, take away those errors, and you're left with 10 or 11 rebounds in your count and 12 rebounds in the box score count. That's a difference of 1 or 2, and that is most certainly not statistically significant. And that's ignoring the fact that you are only doing this for Duke and trying to say that we don't have a rebounding problem relative to other teams.
Your analysis is well-intentioned and I certainly commend the effort. But there are multiple mistakes you've made that render your conclusion invalid.
Hey CDu -
I actually think our areas of disagreement are much smaller than they might seem. Which I'll get to in a minute. But first I want to commend and thank you for the totally friendly and respectful tone of your posts, even though you disagreed with me. It's not always like that on the boards. I like that we can have a frank discussion of these types of issues without anybody getting personal or going off the deep end. That's the way it's supposed to be on these boards. So kudos to you, as well as the others who have posted in the thread.
Unfortunately, some yahoo reading this thread felt it was appropriate to flame me for my last post in this thread, anonymously of course, stating that this discussion we've been having is a "cat fight" of which he disapproves. I hope he didn't do the same to you, but he probably did. There's nothing whatsoever wrong with the content or tone of this thread, as like I said it's been informed and respectful by all involved, and not personal attacks or a "cat fight." Too bad some people can't tell the difference.
In any event, I definitely get your point that the issue for you (and KenPom) is one of relativity, that Duke's defensive rebound numbers, percentages, etc., are not strong relative to other top teams. I don't, and couldn't, disagree with that.
But as I've thought about it more, I think even when viewed in relationship to other teams' numbers, one of the consequences of (as you said you agreed exists) the inflation of the overall raw numbers in the box scores is that the scope of the problem becomes overstated. It's kind of like this: if when you're closely reviewing the tape of a game, if there are -- just picking a number here -- 25% less plays that actually involve a defensive/offensive rebounding opportunity than might appear to be the case when just looking at a box score, then isn't the fact that a team demonstrates even a relative lack of success in grabbing those rebounds a less significant factor in analyzing the team's performance than one would've thought? Boy, that was convoluted. I'm not saying rebounding isn't important. Of course it's very important. But when the true numbers of opportunities are smaller, then the impact of being "bad" at it is diminished. At the extreme, it wouldn't matter so much if you're horrible at defensive rebounding if there aren't any defensive rebounds to grab.
And that's why I think the raw numbers, and getting them right, do matter. It's so funny when you look at the box score(s) and the play-by-play. Just taking the Temple game, I checked a number of sites. ESPN, in its box score, totaled up the Temple players' individual OR numbers and got 14. In the play-by-play, there are 15 times in which an individual player is stated to have gotten an OR. Plus they list 3 offensive rebounds as going to the team. But in the box score they list Watson getting an OR, with DeCosey and Dingle getting none. In the play by play, just the opposite. DeCosey and Dingle have one, but not Watson. Is Pomeroy using ESPN's numbers? Their box score or their play by play? I don't know.
Then on USA Today's site, they have the same boxscore individual numbers as ESPN, but when they add them up, they total them as 16, which the individuals clearly do not add up to. THey add up to 14. USA Today then notes that there were team rebounds too -- but 4 of them, not 2! Were these four offensive, defensive, or what? Were the four included in the 16 somehow? I don't know.
NBC Sports has the same boxscore as ESPN too, but totals them up to 16 like USA Today did, but unlike USA Today, says nothing at all about team rebounds.
WRAL has the same box score as ESPN too, but totals them to 16, but in their case with 2 being attributed to team.
Seems like kind of a mess, and I have no idea whose numbers KenPom is plugging into his formulas.
In my analysis of the plays, I consider the following to be true rebounding plays, the kind that should be counted as real offensive rebounds. I know, my rules, but I just don't think there's any value in counting some of the other plays that were counted in the box score because they evidence nothing about a team's true defensive rebounding prowess or lack thereof. I think the plays at the following times in the game were signficant, true rebounding opportunities that we lost:
the first two that I missed, H-J's at 14:39 and 11:16, Lee's at 14:00, Wyatt's at 8:15, Peppers' at 6:39, Randall's at 11:38 and 9:03 of the second half, DiLeo's at 11:26 of the second half, O'Brien's at 8:19 of the second half, and Dingle's at 0:47 at the end of the game. That's 12, including the garbage time one in the last minute. I would've counted the one that went to team at 11:43 of the first half, because that was a ball Duke could've/should've had, but I thought you said KenPom doesn't count those.
I just don't agree that Lee deserves an OR for his getting a loose ball at 14:39, or that anyone deserves one for the play at 1:36 of the half. Or at 0:39 of the half, where there was no play at all. Or at 2:00 in the second half, which was just a blocked shot by Kelly. That's 4 or 5 then, that got counted as OR's for Temple in either ESPN's box score or its play by play, that really weren't. (4 if the play at 1:36 was one rebound, 5 if counted as 2 -- to me, it was zero)
I don't know how all these numbers could be made to jibe either with each other or with the play by play. All I can do is reiterate that if in a universe of 14 or 16 or 17 offensive rebounds that is being used as a baseline and is being plugged into KenPom's formulas, if 4 or 5 of them really weren't, to me that's a significant number. I get it, other teams' numbers are going to be reduced similarly. And I get that with smaller overall raw numbers, the statistical significance of variation in the numbers actually increases. But nevertheless, with the smaller overall raw numbers, the scope of Duke's problem in this area -- even if it is a little greater than other teams' problem in this area -- is in my analysis not a significant factor in the team's performance. Whew!
Hi tommy,
Sorry to hear you got flamed. I didn't (at least not to my knowledge) and I agree that nothing in this thread warranted that. If anything, I was slightly closer to the line than you, but I felt like we both kept any jabs well above the belt. :) Nice to be able to disagree in a civil manner.
Thanks as well for your followup post. I agree that we aren't that far off. The effort you've gone in this post really helps identify where we diverge. And I think I can help bridge the gap. See below:
Is Pomeroy using ESPN's numbers? Their box score or their play by play? I don't know.
Pomeroy uses box scores, not play-by-play. See the the following, posted by timmy c (took me a moment to realize that this was a different person and that you weren't arguing with yourself!) from Pomeroy's site:
OR% = OR / (OR + DRopp)
Keep in mind that rebounding percentage is computed from box score data which does not contain team rebounds. Therefore, the figures shown here may differ slightly from calculations made on the rebounding totals provided by a school
Based on the "which does not contain team rebounds" I suspect Pomeroy uses ESPN's box scores. In any case, this specifies that (a) he uses box scores and (b) the box scores don't include team rebounds. So you can disregard discrepancies in the play-by-play and box score (rely on the box score) and you can disregard team rebounds.
Basically, he calculates the following: (OR for team A) / (OR for team A + DR for team B), where OR and DR exclude team rebounds.
Then on USA Today's site, they have the same boxscore individual numbers as ESPN, but when they add them up, they total them as 16, which the individuals clearly do not add up to. THey add up to 14. USA Today then notes that there were team rebounds too -- but 4 of them, not 2! Were these four offensive, defensive, or what? Were the four included in the 16 somehow? I don't know.
NBC Sports has the same boxscore as ESPN too, but totals them up to 16 like USA Today did, but unlike USA Today, says nothing at all about team rebounds.
WRAL has the same box score as ESPN too, but totals them to 16, but in their case with 2 being attributed to team.
Seems like kind of a mess, and I have no idea whose numbers KenPom is plugging into his formulas.
ESPN presents 14 individual offensive rebounds and 35 total individual rebounds (meaning 21 individual defensive rebounds). USA Today presents 16 offensive rebounds and 23 defensive rebounds for Temple, with 4 of those being team rebounds. Summing the individual rebounds, you get 14 individual offensive rebounds and 21 individual defensive rebounds, so it's clear that 2 team rebounds were offensive and 2 were defensive. WRAL presents 16 offensive rebounds (2 of those being team rebounds) and 23 defensive rebounds (2 being team rebounds). That leaves 14 individual offensive rebounds and 21 individual defensive rebounds. NBC has 16 offensive and 39 total (23 defensive), but if you count up the individual rebounds it is 14 and 21.
So there's no discrepancy. Each site attributes 14 individual offensive rebounds, 21 individual defensive rebounds. 2 of the 4 explicitly include 4 team rebounds (ESPN does not, while NBC implicitly does), but one can fairly easily determine the number of individual rebounds.
Based on these data, I'm more inclined to believe that Pomeroy's box score of choice is ESPN rather than WRAL or USA, given that he mentions that his box scores don't include team rebounds.
I think the plays at the following times in the game were signficant, true rebounding opportunities that we lost:
the first two that I missed, H-J's at 14:39 and 11:16, Lee's at 14:00, Wyatt's at 8:15, Peppers' at 6:39, Randall's at 11:38 and 9:03 of the second half, DiLeo's at 11:26 of the second half, O'Brien's at 8:19 of the second half, and Dingle's at 0:47 at the end of the game. That's 12, including the garbage time one in the last minute. I would've counted the one that went to team at 11:43 of the first half, because that was a ball Duke could've/should've had, but I thought you said KenPom doesn't count those.
Agreed.
I just don't agree that Lee deserves an OR for his getting a loose ball at 14:39
That's fair.
or that anyone deserves one for the play at 1:36 of the half.
Also fair. Note, though, that DeCosey is not credited with an offensive rebound in the box score anywhere (only the play-by-play on ESPN), so clearly that's just an error in the play-by-play. Since it doesn't appear in the box score, you can disregard it (as Pomeroy disregards it anyway).
Or at 0:39 of the half, where there was no play at all.
That was a team rebound, which (per above) are not included in Pomeroy's stats. So you can disregard this one.
Or at 2:00 in the second half, which was just a blocked shot by Kelly.
Again, disregard, as this is a team rebound.
That's 4 or 5 then, that got counted as OR's for Temple in either ESPN's box score or its play by play, that really weren't. (4 if the play at 1:36 was one rebound, 5 if counted as 2 -- to me, it was zero)
As I've illustrated above, it's a difference of 2, not 4 or 5. So you have 12 of the 14 individual offensive rebounds being attributed as the fault of Duke. That's not a significant difference. And that difference almost certainly comes out in the wash if you look at all games, as random rebound plays like this do happen.
I don't know how all these numbers could be made to jibe either with each other or with the play by play. All I can do is reiterate that if in a universe of 14 or 16 or 17 offensive rebounds that is being used as a baseline and is being plugged into KenPom's formulas, if 4 or 5 of them really weren't, to me that's a significant number.
As noted above, the actual difference is only 2. The play-by-play should only be considered a guide, as I doubt they make much effort to update it after the game, whereas the box scores are more official. It takes some work, but one can weed through the discrepancies if desired. But it's not really necessary (per above).
I get it, other teams' numbers are going to be reduced similarly. And I get that with smaller overall raw numbers, the statistical significance of variation in the numbers actually increases. But nevertheless, with the smaller overall raw numbers, the scope of Duke's problem in this area -- even if it is a little greater than other teams' problem in this area -- is in my analysis not a significant factor in the team's performance. Whew!
The bolded part is an important discussion point. Based on our performance to this point, I agree. And Pomeroy does too, actually. His statement is that, to this point, we have been the following (in no particular order):
1. a pretty poor defensive rebounding team
2. a very efficient offensive team
3. a team good at turning opponents over
4. a team very good at holding team's to low FG%
5. a team very good at winning the free throw battle
As long as we remain elite in the other factors, the poor defensive rebounding may remain a minor point of concern. If, for any reason, we start to regress a bit in the other factors, the rebounds become a red flag.
I'm not saying the rebounds have cost us so far. We're undefeated with some very impressive scalps, so clearly that's not the case. My main point is that I don't think Pomeroy's analysis isn't really overstating anything: he says we're a poor defensive rebounding team (true relative to other teams) but we've been really good at everything else (also true), and that if we stop being elite at the other stuff, the rebounding issue becomes a red flag (in my opinion, also true).
tommy
12-13-2012, 01:04 PM
Hi tommy,
Sorry to hear you got flamed. I didn't (at least not to my knowledge) and I agree that nothing in this thread warranted that. If anything, I was slightly closer to the line than you, but I felt like we both kept any jabs well above the belt. :) Nice to be able to disagree in a civil manner.
Thanks as well for your followup post. I agree that we aren't that far off. The effort you've gone in this post really helps identify where we diverge. And I think I can help bridge the gap. See below:
Pomeroy uses box scores, not play-by-play. See the the following, posted by timmy c (took me a moment to realize that this was a different person and that you weren't arguing with yourself!) from Pomeroy's site:
Based on the "which does not contain team rebounds" I suspect Pomeroy uses ESPN's box scores. In any case, this specifies that (a) he uses box scores and (b) the box scores don't include team rebounds. So you can disregard discrepancies in the play-by-play and box score (rely on the box score) and you can disregard team rebounds.
Basically, he calculates the following: (OR for team A) / (OR for team A + DR for team B), where OR and DR exclude team rebounds.
ESPN presents 14 individual offensive rebounds and 35 total individual rebounds (meaning 21 individual defensive rebounds). USA Today presents 16 offensive rebounds and 23 defensive rebounds for Temple, with 4 of those being team rebounds. Summing the individual rebounds, you get 14 individual offensive rebounds and 21 individual defensive rebounds, so it's clear that 2 team rebounds were offensive and 2 were defensive. WRAL presents 16 offensive rebounds (2 of those being team rebounds) and 23 defensive rebounds (2 being team rebounds). That leaves 14 individual offensive rebounds and 21 individual defensive rebounds. NBC has 16 offensive and 39 total (23 defensive), but if you count up the individual rebounds it is 14 and 21.
So there's no discrepancy. Each site attributes 14 individual offensive rebounds, 21 individual defensive rebounds. 2 of the 4 explicitly include 4 team rebounds (ESPN does not, while NBC implicitly does), but one can fairly easily determine the number of individual rebounds.
Based on these data, I'm more inclined to believe that Pomeroy's box score of choice is ESPN rather than WRAL or USA, given that he mentions that his box scores don't include team rebounds.
Agreed.
That's fair.
Also fair. Note, though, that DeCosey is not credited with an offensive rebound in the box score anywhere (only the play-by-play on ESPN), so clearly that's just an error in the play-by-play. Since it doesn't appear in the box score, you can disregard it (as Pomeroy disregards it anyway).
That was a team rebound, which (per above) are not included in Pomeroy's stats. So you can disregard this one.
Again, disregard, as this is a team rebound.
As I've illustrated above, it's a difference of 2, not 4 or 5. So you have 12 of the 14 individual offensive rebounds being attributed as the fault of Duke. That's not a significant difference. And that difference almost certainly comes out in the wash if you look at all games, as random rebound plays like this do happen.
As noted above, the actual difference is only 2. The play-by-play should only be considered a guide, as I doubt they make much effort to update it after the game, whereas the box scores are more official. It takes some work, but one can weed through the discrepancies if desired. But it's not really necessary (per above).
The bolded part is an important discussion point. Based on our performance to this point, I agree. And Pomeroy does too, actually. His statement is that, to this point, we have been the following (in no particular order):
1. a pretty poor defensive rebounding team
2. a very efficient offensive team
3. a team good at turning opponents over
4. a team very good at holding team's to low FG%
5. a team very good at winning the free throw battle
As long as we remain elite in the other factors, the poor defensive rebounding may remain a minor point of concern. If, for any reason, we start to regress a bit in the other factors, the rebounds become a red flag.
I'm not saying the rebounds have cost us so far. We're undefeated with some very impressive scalps, so clearly that's not the case. My main point is that I don't think Pomeroy's analysis isn't really overstating anything: he says we're a poor defensive rebounding team (true relative to other teams) but we've been really good at everything else (also true), and that if we stop being elite at the other stuff, the rebounding issue becomes a red flag (in my opinion, also true).
I think we have reached a general agreement! Which is nice. . .
toooskies
12-13-2012, 01:05 PM
Having worked in TV for college basketball games, I'm pretty familiar with how the stats are tallied, both according to the official rules as well as how things go scoring in realtime (my company delivers stats in parallel with official stats for some games).
First, background on box scores:
There is a universal Official box score for every single game played in the NCAA. The Official Scorer keeps track of all the relevant stats for the game. It is necessary to do all this because there are some basic rules of the game that need to be true:
- For every shot, there must be points or a rebound. A ball going out of bounds is a team rebound. A first-of-two missed free throw or shot at the end of the half is usually a "dead ball" rebound, which rarely make the box score as they don't signify anything, but they keep the numbers consistent. If there is a shot and the ball stays in play, there MUST be an individual rebound. There is always a rebound after a blocked shot. If the ball falls to the court and rolls 75 feet away before someone secures it, that person gets the rebound. Typically, the rules say a person tipping the ball to someone else should get the rebound, but in practice, unless it's a very controlled tip, the rebound goes to the player who secures the ball.
- Every possession must end in a made basket, a defensive rebound, a turnover, or the clock running out (which I don't believe counts as a possession).
- (Total shots) + (Turnovers) = (individual rebounds) + (team rebounds) + (dead ball rebounds) + (halves starting with the ball) - (halves ending with the ball). This is a statistical verification in the scorebook that the game was run correctly. One team can have up to 2 more possessions than the other team by winning the opening tip and getting the last shot in both halves.
The official statistician creates the official Box Score, and I'm sure KenPom gets a verified copy of this. Typically, printouts of the official Box Score are available to all press. If you see an actual discrepancy between sources for box scores, either they weren't copies of the official one or there was a publishing error.
There is an official play-by-play feed that's been around for several years. I don't know if ESPN broadcasts this directly or if they have their own person duplicating the work. However, I believe KenPom gets the official play-by-play rather than using ESPN's. I do know, however, having duplicated the official play-by-play work for TV, that they don't always get it right, and often not in a timely basis. However it is typically corrected to the final version of the stats at the conclusion of a half. I do not know if the corrections become "official" in the play-by-play feed; I doubt it actually. Which is probably why KenPom uses the Box Score for team ratings of the Four Factors.
To download the official rules, go here. (http://www.ncaapublications.com/p-4227-2011-2013-mens-womens-basketball-rule-book-2-year-publication.aspx) For the statistician's guide book (more relevant), go here. (http://fs.ncaa.org/Docs/stats/Stats_Manuals/Basketball/2013%20Bsk%20Stats%20Manual%20easy%20print.pdf)
toooskies
12-13-2012, 03:22 PM
My main point is that I don't think Pomeroy's analysis isn't really overstating anything: he says we're a poor defensive rebounding team (true relative to other teams) but we've been really good at everything else (also true), and that if we stop being elite at the other stuff, the rebounding issue becomes a red flag (in my opinion, also true).
Regarding this bit of content, it's questionable whether the numbers actually mean we're worse relative to other teams as much as we take and force shots that aren't as likely to be rebounded. As in, we shoot more free throws than opponents (not a bad thing) and we block more shots (not a bad thing).
As a first reference on where shots go, here's an article about the NBA from 82games.com, which is a good first verification of the likelihood of rebounding a given shot, here. (http://www.82games.com/comm13.htm) As you can see, the offensive rebound rate of a free throw is less than half the regular rate. Now, if you shoot better than 60% from the line, then it's a net positive even though you sacrifice offensive rebound rate. Alternately, if you can force a low shooting percentage without fouling, you'll get fewer high-percentage rebounding opportunities. I'm not sure how they came up with these numbers.
KenPom himself took a stab at this in 2004, albeit by extrapolating from box scores. His estimates:
3 point shot: 21.0% (21% of missed 3 pointers are rebounded by the offense)
2 point shot: 41.1%
Free throw: 20.3%
That's from here. (http://kenpom.com/blog/index.php/weblog/entry/rebounding_2s_vs_3s)
I'm a little uncertain that his mathematical solution here is sound, and the fact that they differ from the NBA estimates makes one wonder a little more, but it's perfectly possible that the pro stats differ from college (particularly, that threes in college are different from "jumpers", or alternately, that 2-point jumpers and 3-point jumpers are statistically different).
Also from 82games.com, we have blocked shot rebounding ratios here. (http://www.82games.com/comm16.htm) Again-- we block many more shots than we have our own shots blocked, and those go to the offense about 40% of the time (again in the NBA, there aren't college stats here).
Regarding this bit of content, it's questionable whether the numbers actually mean we're worse relative to other teams as much as we take and force shots that aren't as likely to be rebounded. As in, we shoot more free throws than opponents (not a bad thing) and we block more shots (not a bad thing).
Your point that not all rebounds are the same is very true. However, we're talking specifically about our defensive rebounding only, so the discussion about the type of shots we shoot is irrelevant. Pomeroy isn't saying we're a bad offensive rebounding team. He's saying we're bad at getting defensive rebounds. So it doesn't matter that we don't get a lot of our shots blocked (or that we shoot a lot of free throws) in the grand scheme of things, because those don't really affect our defensive rebounding efficiency.
I don't know where we stand nationally as a shotblocking team (probably above average - perhaps well above average), but I doubt that we block at a rate substantial enough to sway our defensive rebounding efficiency by any meaningful amount (not that many blocks per game anyway, and the difference in rebound rate between blocked shots and unblocked shots isn't substantial enough). I do suspect we're above average at preventing foul shots. How much that affects our defensive rebounding efficiency I can't say. But I doubt it's enough to bump us up from the 200s to a decent range.
tommy
12-14-2012, 08:01 PM
Having worked in TV for college basketball games, I'm pretty familiar with how the stats are tallied, both according to the official rules as well as how things go scoring in realtime (my company delivers stats in parallel with official stats for some games).
First, background on box scores:
There is a universal Official box score for every single game played in the NCAA. The Official Scorer keeps track of all the relevant stats for the game. It is necessary to do all this because there are some basic rules of the game that need to be true:
- For every shot, there must be points or a rebound. A ball going out of bounds is a team rebound. A first-of-two missed free throw or shot at the end of the half is usually a "dead ball" rebound, which rarely make the box score as they don't signify anything, but they keep the numbers consistent. If there is a shot and the ball stays in play, there MUST be an individual rebound. There is always a rebound after a blocked shot. If the ball falls to the court and rolls 75 feet away before someone secures it, that person gets the rebound. Typically, the rules say a person tipping the ball to someone else should get the rebound, but in practice, unless it's a very controlled tip, the rebound goes to the player who secures the ball.
- Every possession must end in a made basket, a defensive rebound, a turnover, or the clock running out (which I don't believe counts as a possession).
- (Total shots) + (Turnovers) = (individual rebounds) + (team rebounds) + (dead ball rebounds) + (halves starting with the ball) - (halves ending with the ball). This is a statistical verification in the scorebook that the game was run correctly. One team can have up to 2 more possessions than the other team by winning the opening tip and getting the last shot in both halves.
Thanks - that's very helpful.
But I gotta say -- and I know you didn't create the rules of course -- but these rules seem to me not just to permit, but to cause, distortions of rebounding numbers. What these rules do is to sacrifice the real accuracy of the numbers, by which I mean having the numbers describe what actually happened in the game, at the altar of neatness and having everything match up and have it tied up in a nice little bow. "Keeping the numbers consistent" as you term it.
"If there is a shot and the ball stays in play, there MUST be an individual rebound."
Why? Sometimes shots happen, and nobody corrals the rebound. It bounces off the basket, it's on the floor, it's a loose ball, and somebody hustles, gets a floor burn, and retrieves it for his team. Sometimes it's 5 to 10 seconds after the ball clangs off the basket that it's actually possessed again. That's not a rebound. It's a loose ball.
"There is always a rebound after a blocked shot. If the ball falls to the court and rolls 75 feet away before someone secures it, that person gets the rebound."
Again, this makes no sense. Somebody picking a ball up on the floor 75 feet away from the basket after someone else blocked a shot and it rolled all that distance? That's no real basketball person's definition of a rebound. This makes it look like one player and one team is better at rebounding than they actually are, and the opponents are worse. Distortion.
Why on earth is it so important to have everything match up and be consistent and the numbers be perfect? It's sports, for crying out loud! It's imperfect and messy on the court. Let it be imperfect and messy on the stat sheet if necessary to have the stat sheet be accurate. Right? Otherwise, what's the point?
Thanks - that's very helpful.
But I gotta say -- and I know you didn't create the rules of course -- but these rules seem to me not just to permit, but to cause, distortions of rebounding numbers. What these rules do is to sacrifice the real accuracy of the numbers, by which I mean having the numbers describe what actually happened in the game, at the altar of neatness and having everything match up and have it tied up in a nice little bow. "Keeping the numbers consistent" as you term it.
"If there is a shot and the ball stays in play, there MUST be an individual rebound."
Why? Sometimes shots happen, and nobody corrals the rebound. It bounces off the basket, it's on the floor, it's a loose ball, and somebody hustles, gets a floor burn, and retrieves it for his team. Sometimes it's 5 to 10 seconds after the ball clangs off the basket that it's actually possessed again. That's not a rebound. It's a loose ball.
"There is always a rebound after a blocked shot. If the ball falls to the court and rolls 75 feet away before someone secures it, that person gets the rebound."
Again, this makes no sense. Somebody picking a ball up on the floor 75 feet away from the basket after someone else blocked a shot and it rolled all that distance? That's no real basketball person's definition of a rebound. This makes it look like one player and one team is better at rebounding than they actually are, and the opponents are worse. Distortion.
Why on earth is it so important to have everything match up and be consistent and the numbers be perfect? It's sports, for crying out loud! It's imperfect and messy on the court. Let it be imperfect and messy on the stat sheet if necessary to have the stat sheet be accurate. Right? Otherwise, what's the point?
Tommy, I hate to disagree with you on the heels of us working so hard to come to an agreement earlier. :) But I do disagree here.
You seem to be applying your own concept of what a rebound is (which is different from the actual definition of a rebound). A rebound is simply any instance in which a player (or team) recovers a missed shot. So rebound statistics accurately reflect which players/teams recover missed shots. They are not intended to be an exact measure of which team boxes out better. They are simply a measure of how many times a player/team recovers a missed shot.
Now, it is true that the vast majority of rebounds fall into the category of rebound that you are comfortable with. As a result, some folks have subcosciously redefined what constitutes a rebound. This does not mean that the rebound stats distort a team's rebounding ability. On the contrary, they exactly reflect a team's rebounding performance. Instead, they merely distort a team's performance in the "redefined" definition of rebounds. But since that's not what the stat is intended to reflect, one should expect that distortion would occur.
Additionally, there just aren't that many examples of these types of rebounds. We found 2 of the 40 rebound opportunities on the Temple offensive end (Temple had 14 individual offensive rebounds, Duke had 26 individual defensive rebounds) in the Temple game were anomalies (remember: we can ignore team rebounds, as that is simply the designation for balls that aren't rebounded by anyone). That's 5%, and I'd guess that's probably on the high side. And I also suspect that these rare plays wash out overall.
With any statistic, there will be random noise. That randomness tends to cancel out over a large sample size, so there isn't really any distortion here (even ignoring that the definition of a rebound necessitates that there is no distortion anyway). And there is no reason to assume that all stats accumulated be done so in a similar way. Not all baskets are acquired equally. Not all 5 yard gains in football are the same. No reason to assume that all rebounds fit into a narrow view of what a rebound is.
Saratoga2
12-15-2012, 11:26 AM
I retired from a business which became obsessed with measuring job performance by developing statistics. I see some similarity in trying to develop a rebounding statistic to determine whether we are doing relatively well compared to other teams. That method can get complicated as this thread clearly indicates, and indeed can obscure what the central needs are to do a better job.
Someone indicated that if we could get a couple of more defensive rebounds a game, it would provide a little more point margin needed against other strong teams. That is a different approach, saying what we need to do and then using experienced based observations to determine how to achieve that goal. Clearly defenders using boxing out more assertively is an area of improvement. In addition, sending in guards to crash the boards might help, although that might result in fast breaks. Another thought is the use of our bigs for long minutes and to run the floor leaves them perhaps so tired they lose some aggressiveness. That along with picking up fouls early can cause our bigs to lack aggressiveness. Having Marshall back should help the bigs maintain their aggressiveness.
I retired from a business which became obsessed with measuring job performance by developing statistics. I see some similarity in trying to develop a rebounding statistic to determine whether we are doing relatively well compared to other teams. That method can get complicated as this thread clearly indicates, and indeed can obscure what the central needs are to do a better job.
It's really not that complicated. It's very simple, in fact: the formula for a team's defensive rebound efficiency is just (OR_teamB)/(OR_teamB + DR_teamA), and the data are the individual rebound numbers from the box scores. It just took a lot of discussion to identify exactly how the method is done, but the method itself is quite simple.
Also, I don't think the approach obscures what is needed to be done to do a better job. All it does is highlight a point of potential concern. It doesn't give any indication (one way or the other) as to what (if anything) needs to happen to address the problem. That's for the coaches to figure out. The stat is merely an indicator of past performance and a "caution flag" for future performance. Nothing more, nothing less. We have not been a good defensive rebounding team, and that's something that (if not addressed) could bite us if we get less awesome at the other stuff.
Someone indicated that if we could get a couple of more defensive rebounds a game, it would provide a little more point margin needed against other strong teams. That is a different approach, saying what we need to do and then using experienced based observations to determine how to achieve that goal. Clearly defenders using boxing out more assertively is an area of improvement. In addition, sending in guards to crash the boards might help, although that might result in fast breaks. Another thought is the use of our bigs for long minutes and to run the floor leaves them perhaps so tired they lose some aggressiveness. That along with picking up fouls early can cause our bigs to lack aggressiveness. Having Marshall back should help the bigs maintain their aggressiveness.
Yes, ultimately what the stats show is that it would be helpful to get a few more defensive rebounds (and correspondingly allow fewer offensive rebounds). As I said above, stats are only intended to show the "what" (as in, what happened), not the "how/why." Figuring out the "how/why" is part of why the coaching staff gets paid the big bucks.
I'm not how sending guards to crash the defensive boards will lead to more fast breaks against us. It might result in fewer of our own fast breaks, but if the ball is in their offensive end already, guard rebounding won't cause fast breaks. Unless you're talking about our offensive rebounding, in which case guard rebounding would open fast break opportunities. But offensive rebounding is not where we're deficient; it's on the defensive glass. I don't think we need to do anything to change our offensive approach.
It does seem that we aren't very good at boxing out. Outside for Hairston, I don't think any of our players do a consistently good job of that. Even Mason, who is a very solid defensive rebounder, usually just relies on his height and athleticism to get boards, often neglecting to really put a body into the offensive player. And none of the guards do a good job of it.
That's a pretty easy thing for the layman to see. Where it gets more complicated is answering "why is that the case?" Is it that the coaching staff isn't good at coaching boxing out (possible, but hopefully not the case)? Is it that our players don't focus on that aspect of defense? Is it that, because we have a limited number of practice hours, that's one of the things that we just don't have time to really focus on in practice (sort of like in bounds plays against pressure defense used to be an issue)? Is it a conscious decision by the coaching staff for whatever reason? I'm inclined to believe it's a combination of lack of time available to focus on this and the corresponding lack of player focus on doing it in the games. But I'm admittedly just speculating here, and I suspect that everyone on DBR is in the same boat (none of us, to my knowledge, have enough access to the coaches, players, and practices to know).
Powered by vBulletin® Version 4.2.5 Copyright © 2024 vBulletin Solutions Inc. All rights reserved.