APBRmetrics

[tarak]

Hi all,

I was trying to come up with an accessible and straightforward measure to identify good outlet passers and/or players who can handle the ball off of defensive rebounds, and get some picture of what effect they have on their team's offense. I played with the play-by-play data and looked at all defensive rebounds from the 07-08 season that immediately preceded a shot, foul, or turnover (so I removed those defensive rebounds where there was a timeout, end of quarter, kicked ball, etc. right after), and then the number of seconds between the rebound and the next offensive play. I also counted the "success rate" where success here is counted as a made shot or a foul called on the other team, as opposed to a missed shot or a turnover (missed shots that result in offensive rebounds are not counted as successes in this study, although I could easily change that) -- these aren't reflections of success just for the rebounder, in an example where a rebounder makes an outlet pass and never is involved in the play after that, he is still credited with a "success" if the team ends up scoring. For individual players, in addition to "average time" I added in a column that compares the average time to offense off of that player's rebounds to the average time to offense off rebounds for the team as a whole -- green means the player's rebounds result in offense faster than average for the team, red means slower. I put up my results as a spreadsheet:

http://spreadsheets.google.com/ccc?key=pnj1rgBpEtJEQCDZf_MqxEg&hl=en

The first two tabs are team and positional averages (the position designations are from dougstats) and then the results for individual players are on the positional tabs. I also wrote about this at my blog a couple of weeks ago (http://fruithoopz.blogspot.com/2009/01/converting-off-of-defensive-rebounds.html) but I don't think that entry really adds anything to what you can find on the spreadsheet. I've done the same process for the 05-06 and 06-07 seasons but didn't publish them (in general, it seemed like the "average time" was similar for players across seasons, although the "success rate" varied). . ..

Anyways, I wanted to see if anyone's done this or something similar before? Or whether there is a way to make this more useful/informative? I know Basketball on Paper mentions a WNBA study of points per play off of offensive rebounds, and I always see the "points off of turnovers" stat on television broadcasts, but I haven't seen anything about offense generated off of defensive rebounds.

Also, is there anything of interest you see in these results? I was struck by how valuable a rebounding point guard can be, for instance.

[Ryan J. Parker]

I'm interested in seeing the data, but I can't access the spreadsheet.
_________________
I am a basketball geek.

[tarak]

Oops, sorry about that. I believe it should be viewable by anybody now.

[Mountain]

The parts of the table are accessible from the box at the article page too.

These "success rates" are hard for me to interpret by themselves. They don't seem that elevated but I don't know for sure. What is the average success rate for all other plays not off a defensive rebound and is it higher or lower overall and team by team? The amount of "running" or early offense influences the average elapsed time but so does the average "pace" of teams half court offense so it is still tough to draw many conclusions without look at all the data doing more comparison with non defensive rebounds initiated offense. Lakers look like the most successful in this data but how much of that is from the fast break and how much from half court offense?

At position level when PGs rebound and just go the play ends the quickest by average. (How often their own shot?) Some bigs accelerate the break with their outlet as well as perimeter guys can but the data suggests many more are really slow about it. Of course it would be ideally helpful to know when the ball got cross half court to help decide who was slow the rebounder/outlet guy or the guys up on the break. And you'd have to check success rate of the break iteself to judge whether being fast or slow was working efficently or not.

[tarak]

Mountain -- you're right, maybe I could make the success rate more clear by contrasting it with the team's average success rate, both after a defensive rebound and in general? That might highlight little snippets like, for instance (I'm just picking randomly) that Tim Duncan's rebounds resulted in a success rate that is 2.5% higher than the success rate after defensive rebounds for the Spurs as a whole. I'll do that this weekend and see how it turns out.

I tried to account for the team pace differences by adding the "team average" time column to the individual player tabs. There are still problems, though -- for instance rebounds by Marcus Camby show up as leading to offense more slowly than rebounds by other Denver players, but he played for an extremely fast team, as seen by the team average column. The fact that he was able to play with a fast-breaking team and get so many defensive rebounds as a big without slowing the team down more signficantly than he did seems advantageous, I think -- like there are only so many centers in the league who can allow the team to get out and run as much as Denver did last year.

Maybe it would be better if instead of just the average time, I break out the time to offense in 0-10 and 10+, to see how often a rebound led to a fast break? I can also try to add in how many of the breaks led to an "own shot" vs. someone else's shot, as you suggested . . .. I'll keep you updated.

[Mountain]

Sounds good.

I largely overlooked the value of the "team average" time column. That was worthwhile, so thanks for the reminder.

[thref23]

I think it would also be interesting to look at offense created by missed shots.

I am not talking about the player pulling down an offensive rebound, but rather the player missing a shot.

[tarak]

thref23, I'm not sure I understand what you're asking. Could you elaborate a little bit? Is it a question about how often a particular player's shot leads to an offensive rebound?

[thref23]

[tarak]

Got it. I don't have a full response, but a couple of quick thoughts that will hopefully be interesting --

1) A while back Eli at counttheasket posted a pretty enlightening chart showing which shot locations lead to offensive rebounds. You can check it out here: http://www.countthebasket.com/blog/2008/03/29/rebounding-by-shot-location/

That doesn't answer your question exactly, but it's an interesting start I think. As expected, and in accordance with the example you gave, shots in the paint are the most likely to lead to offensive rebounds (as well as 3-pointers).

Now that Ryan at basketballgeek has made shot locations available in play-by-plays, I would think it's possible to create similar charts that show which shot locations lead to fast breaks for the other team, which I think would be interesting (do 3-pointers, with those long rebounds, lead to fast breaks for the opponents when they don't lead to offensive boards?).

2) Out of curiosity a little while back I had gone and looked at which players had the highest percentage of their missed shots offensive rebounded. Obviously, the answer is really dependent on the team's offensive rebounding -- as players on good offensive rebounding teams always have more of their misses rebounded by their own team. Also, the data is a little messy because of the way tip-ins and putbacks are recorded in the play-by-play. But roughly, the players most likely to have their misses rebounded by their own team are the players most likely to rebound their own misses themselves (Andris Biedrins, for instance). So I looked at just offensive rebounds by players other than the shooter (again, with the caveat that these numbers aren't exact because I don't think I did a good job of correcting for how tip-ins/putbacks are recorded). The players most likely to have their shots rebounded by a teammate were, as you might expect from the chart at countthebasket, players who shot a lot of threes -- Jarvis Hayes, Daniel Gibson . . .. One guy who stuck out at the top of that list for last year was Willie Green, but Philadelphia was the best offensive rebounding team in the league. Similarly, Charlie Bell is really high on the list, and Milwaukee was the third best OReb% team . . ..

Anyways, hopefully those thoughts are related to your comments . . ..

[cherokee_ACB]

[thref23]

[Chicago76]

It looks like there could be three of four different factors at play here:


Reggie Evans is an above average big at your metric, but is it due to:
-players having faith in his ability to get the board and releasing earlier rather than trying to get the board themself or hanging around on D in case he doesn't get it?
-the coaching philosophy of when to release players to begin the offense?
-Evans' outlet passing abilities?
-Evans' teammates ability in transition?

I don't know of a good way of isolating the factors unless you can segregate the rebounding scenarios based upon:

-assist stats for bigs (more likely to give the big the credit on the outlet if he also posts high assist numbers/ratios overall like a B. Miller). The outlet/half court passing skill isn't necessarily the same, but historically, those we tend to think of as great outlet passers tend to post high assist numbers as well for their positions--Russell, Unseld, Walton, etc.
-rebounding rates for players (more likely to think players release earlier if a player has high rebounding rates)
-coaching philosophy/teammate transition abilities (more likely to credit this possibility if multiple players on the same team tend to post higher than league averages, although you've isolated this by comparing players to team averages).

[Mike G]

This is a fascinating lot of data, which I've just made time to play with. I've taken the liberty of including 'team success % from defensive rebounds' (after converting to .000 form), and a column for 'success difference from team'.

Of players with >60 defensive rebounds last year, those most over their team's success rates:

[tarak]

Thanks for the details Mike G. I have been rather busy with work recently, so haven't had a chance to get back to the data, but I think you've pretty much gone through it better than I would have anyways -- I'll try and post the time breakdowns as <10 and 10+ to get an idea of fast breaks shortly.

Chicago76: I agree with the questions/angles you've posed, they seem like the reasonable avenues to explore once you've seen the numbers. One thing I would take issue with is the claim that Reggie Evans is above average here -- while it's true that his time and success% numbers are better than average for his position, as you and others have pointed out, those numbers are really contextual and so the better comparison is probably to one's own team, and by that measure, Reggie Evans doesn't look above average....

I'm intrigued by the idea of a player's individual rebounding rate affecting the rest of the team's decision to release and how that affects these numbers.