Browse Archives by:

Revisiting The Offensive Rebounding Debate

The CelticsHub staff is in Boston this weekend for the 2013 Sloan Sports Analytics Conference. Yesterday, Michael Pina and I discussed the results of a panel discussion and research paper called “To Crash or Not To Crash: A quantitative look at the relationship between offensive rebounding and transition defense in the NBA,” by MIT’s Jenna Wiens, Guha Balakrishnan, Joel Brooks and John Guttag.

You can read our back-and-forth in full over at True Hoop, but here’s one of the excerpts that touches on the ongoing discussion about Boston’s aversion to crashing the boards.

Pina: Just look at this: Three of the NBA’s top four defensive teams (Bulls, Pacers and Grizzlies) also rank in the top five for offensive-rebounding percentage, per NBA.com/Stats. The other defensive team is the Spurs. And where does their offensive rebounding rank? Dead last. Yikes.

DeGama: There’s an interesting bit of overlap between Chicago and Boston but it comes with wildly different results. Under Tom Thibodeau, Chicago has turned out the same kind of sexy defensive efficiencies his Boston teams did while he was working under Doc Rivers. Are we to believe Thibs has changed his perspective on offensive rebounding, that he never held the kind of antipathy toward it Rivers seems to, or are Chicago’s offensive rebounding feats just a function of his roster?

Pina: I’d be more inclined to believe it’s a function of having bears in the frontcourt. Taj Gibson, Carlos Boozer (contract be damned, the man can board), All-Star center Joakim Noah, and — not this season, but in the past two — Omer Asik are superior to Kevin Garnett and a couple hobbled co-stars. The Bulls crash the offensive glass and hold down one of the league’s best defenses. If they could only score the ball, they’d have it all.

DeGama: I think the actionable insights in this area will come when we can account for those roster differences. Speaking of which, this November 2012 quote from Rivers makes it sound like the Celtics have rock-solid proprietary metrics that speak to their specific roster: “That is a number I rarely look at, offensive rebounds. Statistically, it holds up. I can tell you, you don’t offensive rebound, you stop [the opponent’s] transition, you win more games than when you get offensive rebounds. I can guarantee you that on those stats.”

You can read the entire research paper at the Sloan Sports website right here.

  • Phil725

    If we assume the Celtics' season long defense numbers put them in the top 5 soon (which they will, since they're just being dragged down by a bad start; they've been great since, and they're only .1 points behind the Clips,) we'll have a top 5 defense list that includes the two worst offensive rebounders in addition to those three who are great at it. Hard to do results based analysis off that…

    I share the interest of you guys here. I've disagreed with the 'get everyone back' philosophy for a while now, and it's good to see the answers getting closer. I assume the ultimate finding will be that it's team based, with totally ignoring offensive rebounds being a mistake due to law of diminishing returns, but I'm fine with waiting for the smart guys to figure that out.

    • check12check

      haven't read your second post, but i could not agree more with what you said here. IMHO when the math guys figure it out, they will see that the ideal situation is letting a few guys who are close to the glass try to get boards while perhaps ideally 3 guys get back to stop a break in the event that the defense pulls down the board. i know that sounds really overly simplistic, but it A. makes the most sense logically and B. seems to pass the eye test to me.

      one of the many things i love about this sport is that it somehow magically straddles the line between logic/eye test and math/ advanced metrics. basketball is somehow a very simple game one hand, and (I believe at least) the most complicated game out there to "figure out"

      • Phil

        I couldn't agree more about the combination of eye test and stats. I can watch every game and follow the action, then look back on the stats later and confirm or challenge that knowledge. I feel like I know basketball better and better the longer I'm a fan because of it. To me, the stats don't take away the magic, they just allow me to better understand the intricacies of what I'm watching, and I love that.

        I do disagree with basketball being the most complicated though; we know nothing about football from a statistic perspective. That analytic movement is decades behind the other sports because every play is so complex. Basketball is five guys trying to do one thing (score or defend.) Football is eleven guys, all with vastly different jobs, and ten guys could do their jobs perfectly and have everything ruined by one guy, and you may not even know who screwed up based on not knowing who had what responsibility. It actually takes some enjoyment away for me, since I know no matter how much work I put into understanding it, it's going to be closer to movie entertainment for a long time. I just have to sit back and watch.

        For offensive rebounding though, I'd love to see numbers on the difference between four guys back and five guys back. I'd be shocked if they were anything substantial at all, especially once you consider that the other team is going to send at least two guys to rebound if you send one. No one's going to attack 3 on 4, and it's welcomed if they do. I'm to the point where I think that sending back all five to defend transition may hurt transition D, since you can go right from a rebound to attacking immediately. Sending one guy slows that down, and even scoring one or two extra possessions is huge.

  • skeeds

    I never gave much attention to this argument. I think there's nothing to dispute. Offensive rebounding is very useful, when your team CAN REBOUND. It's not a goal in itself, it's rather a very useful way to stop transition offense and get more offensive possessions. Thus, for teams like the Clips and Nuggets, with monster frontcourts featuring the league's best athletic big men (Blake, DeAndre, McGee, Faried), not crashing the boards would be stupid because it's their best weapon when it comes to stopping the opposing offense. For teams like the Spurs and the Celtics, where the physical ability to be productive on the offensive glass simply isn't there, fighting for the offensive rebound you probably won't get anyway, takes precious time out of getting back on defense.
    It's like arguing that blocking shots has nothing to do with good defense because guys like Tim Duncan and KG can play great defense without averaging 4 blocks a game. If you're KG, it might not matter, if you're Dwight Howard though it's your best defensive weapon.

    • check12check

      one minor amendment. Timmy D historically has a pretty solid block stat with 2.2 a game including 2.9 a game one season. Also, he has also had some monstrous block totals this year (don't know what he's averaging but I am a HUGE fan so i check his stat lines quite a bit).

      KG's #s are less sexy. ,

      • check12check

        sorry i used also like that twice in one sentence. I am a horrible person and deserve pain and suffering

  • Anthony

    I think too many times these advanced stats/ metrics are over-emphasized. One simple fact that seemed to be overlooked is that the teams with high off reb % CAN'T SHOOT. MEM, CHI, IND all have poor shooting efg%.

    Three of the worse off reb team (SA, BOS, MIA) have the very high efg%. BOS is a bit of an anomaly because they take so many long 2pt fgs and Doc has always preferred to get back in transition as opposed to chasing the off reb. Traditionally, at least in the KG era, BOS always rank as one of the best efg% teams.

    What about teams that have low off reb % and low efg %? They just aren't very good. Those teams include the Wizards, Bobcats, 76ers, Suns, Raptors.

    I like advanced stats/ metrics as much as the next person but it's much more useful for a game like baseball rather than basketball IMO.

    • Phil

      Little problems like that are why the paper openly states that it's going to take years of quantitative analysis before we have real answers. Analytics work well in baseball because it's so simple; a pitcher throws the ball and the batter hits it. Besides some slight hiccups with fielding, it's cut and dry done. Basketball is more complex than that since it's five on five at all times, but it doesn't mean analytics are useless. We know a ton already that we didn't before (3s and FTs make up a good offense, stuff like that,) and we're getting better and better at figuring out what wins games from a team perspective.

      A huge part of analytics is understanding what you don't know though. No one has all the answers, and no one's trying to say they do. But you have to start somewhere, and in a few years maybe we'll have concrete data on stuff like the ideal offensive rebounding commitment. No one knows that just from watching the game, so where does delving into numbers hurt?

  • mmmmm

    Several of the comments here are struggling to resolve the fact that a few of the best defensive teams are highly ranked in ORB% while all the rest really suck at ORB%. The resolution of this isn't to say that there isn't enough data or that basketball can't be adequately understood with statistics. The reason is simply that ORB%, in general doesn't correlate with OR anti-correlate with defense. For that matter it doesn't, in general, correlate or anti-correlate all that well with much of anything important.

    Important words chosen there: "in general" and both "correlate" as well as "anti-correlate". I'm not saying here that ORBs don't matter. I'm just saying that being good at them doesn't tend to correlate well over time with success or failure (winning & losing games) or even with more direct indicators of success (offensive efficiency).

    If you look beyond this one single 2/3 of a season, the same quandary folks are talking about here is exhibited in almost every season going back for a long, long time. I've looked back as far back into the early 80s and I see no correlation between ORB% and success. Nor with failure.

    I strongly suspect that an analysis that measured the share of close-to-the-rim shots that a team takes probably correlates with ORB% on good defensive teams. Basically, when you have offensive players who play closer to the rim, you are more likely to have players in 'natural position' for offensive rebounds _without_ the need to 'crash'. I think this is probably the most obvious difference between Chicago and Boston. I don't believe Thibodeau really has his team 'crash' the boards more. Players like Noah and Boozer, though, simply play offense closer to the basket than Garnett and Bass.

    There is some simply math that supports Doc Rivers' stated strategy.

    From the moment the ball leaves the hand of one of your players when he takes the shot, your goal is to get the ball back before the other team scores.

    If you crash the boards, the absolute best teams' at ORB% tend to get the ball back at around 32%. So that's basically your best chance to get the ball back before they score if you go that route.

    Alternatively, you can concede them all but the most easy ORBs and instead get back and play defense. If you can prevent easy transition buckets, keep them out of the paint and hold them to a very low FG%, say ~42% on the first shot of their possession, then they've got a 58% chance of missing their first shot. If you can grab that at even a mediocre league-average DRB% rate of ~73%, then you've basically got about a 42% chance of getting the ball back in your hands before the other team scores by going this route.

    All these numbers are variable, but 42% >>> 32%. And this presumes the _best_ ORB% rate for route 1.

    Even if we assume you are the _worst_ defensive rebounding team (70% this season) you still come out with around a 41% chance.

    Obviously, any such strategy has to be considered just the 'general' strategy and there are times, such as end-of-game possessions where you would not want to follow it.

    And, of course, if you are incapable of playing such tight defense so as to hold the other team to a low 40s FG%, this strategy fails.

    The Celtics, early in the season, struggled with that very point. With so many new players, they had trouble executing their defense and opponents were routinely scoring at a high rate inside the paint. Further, though this was more bad luck, teams were hitting long 3s against the C's at a ridiculous high rate. This lead to their opponent eFG% to be over 50% for much of the early part of the season.

    The Celtics opponent FG% rates have fallen steadily since late December (right around 42% since then) and the DRtg of the Celtics has correspondingly gotten better as well. A lot of this is, of course, the return of Bradley, but the team in general is simply executing the defense better – even when Bradley isn't on the floor.

    To return to whether ORBs are good are bad. It think it is unquestioned in my mind that they are good if you get them. But the math supports Rivers' contention that you don't want to go out of your way to prioritize them too highly. I also think that it is in general good for an offense to have an inside presence – a low-post game – instead of relying too heavily on jump shots. And that if you have the personnel for an inside game, that you will tend to get ORBs naturally, without having to change your defensive priorities.

    Of course, if you simply make your shots, ORBs dramatically decrease in importance.

    • hydrofluoric

      Very cool analysis, but a few items stick out to me…

      1. The opposing team here is assumed to miss a shot on 58% of their offensive possessions. However, this doesn't account for free throws or turnovers. I'd like to see a similarly simple analysis that incorporates the "true" likelihood of getting the ball back without the opposing team's scoring.

      2. Corralling an offensive rebound is more valuable than corralling a defensive rebound and then going for it. I don't have the stats on hand but when a team gets an offensive rebound, it often leaves the defense badly out of position. A more complete analysis might look at the result of the ensuing offensive possession as well.

      Maybe I should try doing some research… now I'm interested…

      • mmmmm

        On (1): The numbers up above are 'back of the envelope'. But corrections such as free throws and turnovers are secondary effects. The above calculation is based on the primary event of the first shot of a possession. A defensive foul that leads to completed free throws obviously hurts the strategy (because they got points before you got the ball) but turnovers work to support the strategy (because you definitely get the ball back before they scored). There are small effects that work pro-and-con around it, but the basic math is solid. As pointed out, even if you assume you _could_ grab ORBs at a league-leading rate and were only good enough to grab DRBs at the league worst rate, so long as you can suppress the opponent FG%, the strategy has a very healthy margin.

        On (2): The game value of an offensive rebound is that it preserves an offensive possession. From that point, there is an _chance_ somewhere between 0% and 100% that the team will still get points from the possession. Obviously very high on an at-the-rim put-back and not so high on a long rebound near the 3PT line.

        Meanwhile, the game value of a defensive rebound is that it 100% ends the chance of the other team from scoring on that possession. Thus it results in the offensive team getting exactly 0 points.

        Both are good things to get. The issue here isn't necessarily which is 'more valuable'. The issue is whether to prioritize getting ORBs versus other things. It is possible to say, 'getting an ORB is valuable', but it may or may not be worth the effort of crashing the boards to try to get them – because you have to figure on: What are your chances for success? If you crash the boards (and the referenced SLOAN article shows that you really need at least 3 bodies committed to make the chances even get up into the mid-30s) then you are potentially compromising your transition defense. So if you commit for it but don't grab it, you may be giving up more points on the subsequent defensive possession.

        As pointed out, these are, of course, just general trends and characteristics. Particular situations and players may not make this the best strategy.