Relationship between OFFENSIVE REBOUNDING, ORTg and DRTg

[homecourtloss]

It’s interesting to see the relationship between offensive rebounding and ORTg and DRTg. We know that OREB% has gone down as three point attempt rate has increased.



There was discussion in the mid and late 2000s about the tradeoff between scoring via offensive rebounding vs. what you give up on defense presumable as you’d not be able to get back as effectively if actively attempting to offensive rebound. I believe Popovich discussed this.

I found the correlation coefficient between OREB% and ORTG for every season between 1974 and 2025. The closer to 1.0 you get means that for that particular season, higher offensive rebounding% correlated more strongly with higher ORTg.

I also found the correlation coefficient between OREB% and DRTG for every season between 1974 and 2025. The closer you get to 1.0 means that higher offensive rebounding % correlated with a HIGHER DRTG (i.e., higher offensive rebounding% correlated with a worse defense) and the closer you get to -1.0, means that higher offensive rebounding% correlated with a LOWER DRTG (i.e., higher offensive rebounding% correlated with a better defense).

Most of the r values are modest and correlation doesn’t mean causation, but it’s interesting to see the trends in the NBA through the various eras.



You can also see them both at the same time. You can see in most seasons, higher offensive rebounding % correlated with higher ORTG as one would think but there are seasons when it also correlated with higher DRTG. When the the blue line is inside red line, higher offensive rebounding % correlated more strongly with higher defensive ratings than it did with higher offensive ratings. When you have a big gap, it means that higher offensive rebounding correlated with higher ORTGs and lower DRTGs. Before the 3pt era and from the late 1980s to the latter part of the 1990s, offensive rebounding correlated with higher ORTGs and lower DRTGs. This was also the era when higher ORTGs correlated the strongest with LOWER DRTGs.



It’s also interesting to see the relationship through the years of the correlation between ORTG and DRTG, i.e., what if any relationship does higher ORTGs have on defense when looking at the league as a while.



You can see all the seasons here

Spoiler:

[penbeast0]

Really enjoyed reading this. I don't have much to add but just wanted to thank you for posting it.

[NO-KG-AI]

[instagram][/instagram]

Really enjoyed reading this. I don't have much to add but just wanted to thank you for posting it.

Big agree.

Wonder if correlations can be made to great bigs. Great bigs usually have big impact in both places, rebounding and defense.

Having to commit extra bodies to keeping w dominant big off the offensive glass probably effects transition offense. Likewise, great bigs usually effect defense greatly even without the addition of the above point.

Again, great stuff, and this has always been a point of philosophy and a delicate give and take.

[ReggiesKnicks]

This data does a good job of displaying paradigm shifts in the NBA. I'll avoid making grandiose assumptions about why the shifts occurred and instead reference them. Right around the middle of the 1990s there is a jump in correlation, which coincidentally is in 1994, where the NBA shifted the 3P line forward.

Another interesting point is from 1990-1995, there was a gigantic difference between the good teams in the bad teams. Why? Because there is no reason teams who crash the glass more often (Higher OREB%) are also better defensive teams. There is logically a trade-off when putting pressure on the Offensive Glass, and that is sacrificing defense, notably fast-break and open-court, which leads to high efficiency offense. For the good teams in the period mentioned above, 1990-1995, it doesn't appear that trade-off occurred.

Whether the half-decade was due to talent, coaching, or a mixture of both, it is interesting, nonetheless.

2022 is another interesting data point.

[SHAQ32]

So, basically, more offensive rebounds = more points = less transition/fastbreak opportunities for the opponent?

[Djoker]

The opportunity cost of offensive rebounding depends on the tactical approach of the opponent. If you're playing a team that is consistently slow at pushing the ball up the floor, then you can not only get away with it but benefit significantly. But if you're playing a team determined the push the pace, then it's a very dangerous strategy. You're going to find yourself outnumbered in front of your own basket and typically without any bigs since they are the ones on the offensive glass and lagging on the play.

It's funny because I'm literally watching the 1982 WCF. The Spurs are an elite offensive rebounding team as the commentators point out over and over. But maybe the Showtime Lakers aren't the best team to attack the offensive glass against. Needless to say, the Lakers pushed the pace like crazy and just killed the Spurs on the break in this series. Hell the Lakers were pushing the pace even off of makes. Corzine grabs an offensive board and scores a lay up but Magic, Nixon and Cooper are already past half court. Dunk or lay up within 3 seconds of the Corzine putback. And it never occurred to Stan Albeck to maybe tell his boys to ease off the offensive glass and make sure they've got people back.

[homecourtloss]

This data does a good job of displaying paradigm shifts in the NBA. I'll avoid making grandiose assumptions about why the shifts occurred and instead reference them. Right around the middle of the 1990s there is a jump in correlation, which coincidentally is in 1994, where the NBA shifted the 3P line forward.

Another interesting point is from 1990-1995, there was a gigantic difference between the good teams in the bad teams. Why? Because there is no reason teams who crash the glass more often (Higher OREB%) are also better defensive teams. There is logically a trade-off when putting pressure on the Offensive Glass, and that is sacrificing defense, notably fast-break and open-court, which leads to high efficiency offense. For the good teams in the period mentioned above, 1990-1995, it doesn't appear that trade-off occurred.

Whether the half-decade was due to talent, coaching, or a mixture of both, it is interesting, nonetheless.

2022 is another interesting data point.

It is interesting that in that 1990-1994 period, you see the correlation coefficient between DRTG and OREB% hit -.2 for the first time and staying in that range and even hitting -.4 in 1994 while also keeping a strong (relative to this data set) correlation with ORTG and then a shift happens. As soon as 1997, the r between OREB% and ORTG has basically gone down to 0 and then in the early 2000s, the r between OREB% and DRTG goes up to .385 by 2006. The periods of 1990-1994 and 2000-2006 see drastic shifts. And, as you mentioned, 2022 is interesting that it’s the first year since 1994 that we see a -.2 r between OREB% and DRTG and the highest r between OREB% and ORTG since 1994 and 1986.

[homecourtloss]

So, basically, more offensive rebounds = more points = less transition/fastbreak opportunities for the opponent?

Yes, but the interesting thing is that

1) As an entire league, higher OREB% didn’t always correlate with higher ORTGs. In 18 seasons, the r was <.1 with the 2000 season at -.189, 2007 at -.141, 2020 at -.123, 2023 at -.148.

2) As an entire league, higher OREB% correlated with higher DRTGs (i.e., more points allowed) in 17 seasons.

3) in 15 seasons, higher OREB% correlated with a net loss tradeoff, i.e., OREB% correlated higher with higher DRTGs than it did it with ORTG with some years seeing higher OREB% correlate with lower ORTGs AND higher DRTGs. Look at most seasons between 2000- 2007, 2011, 2020, 2023.

The opportunity cost of offensive rebounding depends on the tactical approach of the opponent. If you're playing a team that is consistently slow at pushing the ball up the floor, then you can not only get away with it but benefit significantly. But if you're playing a team determined the push the pace, then it's a very dangerous strategy. You're going to find yourself outnumbered in front of your own basket and typically without any bigs since they are the ones on the offensive glass and lagging on the play.

It's funny because I'm literally watching the 1982 WCF. The Spurs are an elite offensive rebounding team as the commentators point out over and over. But maybe the Showtime Lakers aren't the best team to attack the offensive glass against. Needless to say, the Lakers pushed the pace like crazy and just killed the Spurs on the break in this series. Hell the Lakers were pushing the pace even off of makes. Corzine grabs an offensive board and scores a lay up but Magic, Nixon and Cooper are already past half court. Dunk or lay up within 3 seconds of the Corzine putback. And it never occurred to Stan Albeck to maybe tell his boys to ease off the offensive glass and make sure they've got people back.

The bolded is a given, but I was looking at league-wide trends. In that era, 1981-1983, OREB% correlated strongly with higher ORTGs relative to the dataset.

Spoiler:

[Djoker]

The opportunity cost of offensive rebounding depends on the tactical approach of the opponent. If you're playing a team that is consistently slow at pushing the ball up the floor, then you can not only get away with it but benefit significantly. But if you're playing a team determined the push the pace, then it's a very dangerous strategy. You're going to find yourself outnumbered in front of your own basket and typically without any bigs since they are the ones on the offensive glass and lagging on the play.

It's funny because I'm literally watching the 1982 WCF. The Spurs are an elite offensive rebounding team as the commentators point out over and over. But maybe the Showtime Lakers aren't the best team to attack the offensive glass against. Needless to say, the Lakers pushed the pace like crazy and just killed the Spurs on the break in this series. Hell the Lakers were pushing the pace even off of makes. Corzine grabs an offensive board and scores a lay up but Magic, Nixon and Cooper are already past half court. Dunk or lay up within 3 seconds of the Corzine putback. And it never occurred to Stan Albeck to maybe tell his boys to ease off the offensive glass and make sure they've got people back.

The bolded is a given, but I was looking at league-wide trends. In that era, 1981-1983, OREB% correlated strongly with higher ORTGs relative to the dataset.

Spoiler:

Correlations of 0.3 to 0.6 are not what one would call strong correlation. There is some correlation but honestly given the way those graphs zig-zag I'd probably say ORB and ORtg have a weak correlation until the mid-90's and then no correlation since while ORB and DRtg have no correlation for the entire span. Which I guess intuitively makes sense because offensive rebounding is a tactical approach with a clear opportunity cost and there would be a lot of variance depending on opposing matchups. And as the game became perimeter oriented, it became easier to counter offensive rebounding teams by simply killing them on the break.

That said, your analysis is super interesting!

[homecourtloss]

The opportunity cost of offensive rebounding depends on the tactical approach of the opponent. If you're playing a team that is consistently slow at pushing the ball up the floor, then you can not only get away with it but benefit significantly. But if you're playing a team determined the push the pace, then it's a very dangerous strategy. You're going to find yourself outnumbered in front of your own basket and typically without any bigs since they are the ones on the offensive glass and lagging on the play.

It's funny because I'm literally watching the 1982 WCF. The Spurs are an elite offensive rebounding team as the commentators point out over and over. But maybe the Showtime Lakers aren't the best team to attack the offensive glass against. Needless to say, the Lakers pushed the pace like crazy and just killed the Spurs on the break in this series. Hell the Lakers were pushing the pace even off of makes. Corzine grabs an offensive board and scores a lay up but Magic, Nixon and Cooper are already past half court. Dunk or lay up within 3 seconds of the Corzine putback. And it never occurred to Stan Albeck to maybe tell his boys to ease off the offensive glass and make sure they've got people back.

The bolded is a given, but I was looking at league-wide trends. In that era, 1981-1983, OREB% correlated strongly with higher ORTGs relative to the dataset.

Spoiler:

Correlations of 0.3 to 0.6 are not what one would call strong correlation. There is some correlation but honestly given the way those graphs zig-zag I'd probably say ORB and ORtg have a weak correlation until the mid-90's and then no correlation since while ORB and DRtg have no correlation for the entire span. Which I guess intuitively makes sense because offensive rebounding is a tactical approach with a clear opportunity cost and there would be a lot of variance depending on opposing matchups. And as the game became perimeter oriented, it became easier to counter offensive rebounding teams by simply killing them on the break.

That said, your analysis is super interesting!

Regarding the bolded, in general, no, but looking at the data set and seeing which individual factors correlate with ORTG, an r of .4 actually stands out because, as one would expect, no one individual factor will correlate very strongly with ORTG unless it’s eFG or TS. So when you see an r of .4 or higher the relationship is noticeable, even though it might not necessarily be deterministic.

If you have an r of .4 then the r^2 value is .16. as you know, r^2 is the coefficient of determination and tells us how much of the variance in the dependent variable, i.e., ORTG, can be explained by the independent variable, i.e., OREB%. If r^2 is .16, it means 16% of the variation in ORTG across teams in that season can be accounted for by OREB%. The other 84% of the variation can be explained by other factors, e.g., FTr, 3PAr, TOV%, etc. And this makes sense—you wouldn’t expect any one factor to determine the variance at like .75 or higher in ORTG unless that’s eFG or TS%, which actually have multiple components to them. If you run a test of statistical significance based in a 30 team lesgue, any r of .35 or higher or -.35 or lower will be statistically significant at the usual 5% significance level. And, depending on what we are considering the dataset, the r could be lower and still be statistically significant.

I actually looked at relationship between OREB%, TOV%, FTr, pace, and some other factors and ORTG (I’ll post those some time later) but an r of .4 or even .35 stands out for any season because it doesn’t happen very often. For example, what do you suppose was the highest r between free-throw rate and ORTG was?

.572 in 1998
.524 in 2013
.508 in 1992
.493 in 2018

Or something like pace which an interesting one.

[trex_8063]

You can also see them both at the same time. You can see in most seasons, higher offensive rebounding % correlated with higher ORTG as one would think but there are seasons when it also correlated with higher DRTG. When the the blue line is inside red line, higher offensive rebounding % correlated more strongly with higher defensive ratings than it did with higher offensive ratings. When you have a big gap, it means that higher offensive rebounding correlated with higher ORTGs and lower DRTGs. Before the 3pt era and from the late 1980s to the latter part of the 1990s, offensive rebounding correlated with higher ORTGs and lower DRTGs. This was also the era when higher ORTGs correlated the strongest with LOWER DRTGs.

That massive gap right around '95 is interesting. Any ideas on why the massive split there?

The only thing I can come up with is that's when they shortened the 3pt line. So suddenly you had teams spamming WAY more three's than they had just a year prior, but they had NOT adjusted their 'crash the glass vs get back defensively' strategy/model to compensate for the change.

The period from the LATE 90's through the mid-late 00's is also interesting for the degree of parity: the gains increased OREB% produces in ORtg seems to be almost precisely mitigated by the diminishing returns in DRtg for about 8-9 years there.

[Djoker]

The bolded is a given, but I was looking at league-wide trends. In that era, 1981-1983, OREB% correlated strongly with higher ORTGs relative to the dataset.

Spoiler:

Correlations of 0.3 to 0.6 are not what one would call strong correlation. There is some correlation but honestly given the way those graphs zig-zag I'd probably say ORB and ORtg have a weak correlation until the mid-90's and then no correlation since while ORB and DRtg have no correlation for the entire span. Which I guess intuitively makes sense because offensive rebounding is a tactical approach with a clear opportunity cost and there would be a lot of variance depending on opposing matchups. And as the game became perimeter oriented, it became easier to counter offensive rebounding teams by simply killing them on the break.

That said, your analysis is super interesting!

Regarding the bolded, in general, no, but looking at the data set and seeing which individual factors correlate with ORTG, an r of .4 actually stands out because, as one would expect, no one individual factor will correlate very strongly with ORTG unless it’s eFG or TS. So when you see an r of .4 or higher the relationship is noticeable, even though it might not necessarily be deterministic.

If you have an r of .4 then the r^2 value is .16. as you know, r^2 is the coefficient of determination and tells us how much of the variance in the dependent variable, i.e., ORTG, can be explained by the independent variable, i.e., OREB%. If r^2 is .16, it means 16% of the variation in ORTG across teams in that season can be accounted for by OREB%. The other 84% of the variation can be explained by other factors, e.g., FTr, 3PAr, TOV%, etc. And this makes sense—you wouldn’t expect any one factor to determine the variance at like .75 or higher in ORTG unless that’s eFG or TS%, which actually have multiple components to them. If you run a test of statistical significance based in a 30 team lesgue, any r of .35 or higher or -.35 or lower will be statistically significant at the usual 5% significance level. And, depending on what we are considering the dataset, the r could be lower and still be statistically significant.

I actually looked at relationship between OREB%, TOV%, FTr, pace, and some other factors and ORTG (I’ll post those some time later) but an r of .4 or even .35 stands out for any season because it doesn’t happen very often. For example, what do you suppose was the highest r between free-throw rate and ORTG was?

.572 in 1998
.524 in 2013
.508 in 1992
.493 in 2018

Or something like pace which an interesting one.

What you're saying makes sense.

I would like to see the correlation between OREB and Net Rtg as well. That will really give us a clue about the net effect of offensive rebounding.

[lessthanjake]

So the correlation between ORB% and ORTG in the earlier years is relatively weak in an absolute sense but also pretty significant in a context where there’s so many other factors at play. Then the correlation basically goes to close to zero.

My guess is that the *offensive* opportunity cost of offensive rebounding actually changed a fair bit, because differences in ORB% started being more about teams opting for different spacing strategies, rather than just differences in the rebounding prowess of their players. Specifically, in the late 1990s and in the 2000s, you started to see much more of the stretch 4. Not that it didn’t exist at all before, but it was much less common—with teams mostly having opted for two conventional big men hanging out relatively close to the basket. Obviously, it took a while for this to get to the logical place of having big men widely shooting a lot of threes. But, even before the modern three-point-heavy era, this made there be a real strategic tradeoff offensively. You could stretch the floor and improve spacing by having big men shooting jump shots, but that’d naturally result in getting fewer offensive rebounds, since you’d have fewer guys near the basket. We now know that stretching the floor was probably actually *better* so it’s not surprising to see the correlation go down in later years, as some teams started moving to strategies that emphasized spacing over offensive rebounding. We’d expect those strategic decisions to result in those teams having a lower ORB% without really having a lower ORTG. In contrast, prior to that, most of the difference in team ORB% was probably less a result of strategic choice (since teams were relatively homogenous in not spacing the floor much with their PF and C), and more just a result of having big men who were better at rebounding (which, obviously, is on its own very good for ORTG).

Basically, if both teams have a PF and C staying near the basket, then your PF and C being better offensive rebounders will likely result in having a better ORTG. In contrast, if one team has a PF and C near the basket and the other has a C near the basket but uses the PF to stretch the floor, the first team will likely have a higher ORB% but we might actually even expect the second team to have a higher ORTG because of the power of spacing. In the first scenario, the teams with better ORB% don’t really have a tradeoff for that, while in the second scenario they do. So we’d expect a higher correlation between ORB% and ORTG in the first scenario. I think the league was more like the first scenario in the early set of years in these graphs, and then moved more towards having a lot more of the second scenario. And I see this data as being fairly consistent with that picture.

I’ll note that I think the above goes to the point that we should always be careful to not conflate data on correlations at a team level with correlation/impact at a player level. At a team level, a higher ORB% might not correlate very highly with ORTG, because it implicates strategic tradeoffs that the team is making. But, at a player level, you do still want a player who gets more offensive rebounds. So, for instance, I wouldn’t look at this data and say that it doesn’t matter that Steven Adams is way better at offensive rebounding than some other non-floor-spacing big man. That definitely matters a lot and is really impactful. Basically, if the team-level correlation isn’t very big due to strategic tradeoffs, it’s still the case that holding strategic decisions constant, you still get a lot of value from a great offensive rebounder.

[homecourtloss]

Correlations of 0.3 to 0.6 are not what one would call strong correlation. There is some correlation but honestly given the way those graphs zig-zag I'd probably say ORB and ORtg have a weak correlation until the mid-90's and then no correlation since while ORB and DRtg have no correlation for the entire span. Which I guess intuitively makes sense because offensive rebounding is a tactical approach with a clear opportunity cost and there would be a lot of variance depending on opposing matchups. And as the game became perimeter oriented, it became easier to counter offensive rebounding teams by simply killing them on the break.

That said, your analysis is super interesting!

Regarding the bolded, in general, no, but looking at the data set and seeing which individual factors correlate with ORTG, an r of .4 actually stands out because, as one would expect, no one individual factor will correlate very strongly with ORTG unless it’s eFG or TS. So when you see an r of .4 or higher the relationship is noticeable, even though it might not necessarily be deterministic.

If you have an r of .4 then the r^2 value is .16. as you know, r^2 is the coefficient of determination and tells us how much of the variance in the dependent variable, i.e., ORTG, can be explained by the independent variable, i.e., OREB%. If r^2 is .16, it means 16% of the variation in ORTG across teams in that season can be accounted for by OREB%. The other 84% of the variation can be explained by other factors, e.g., FTr, 3PAr, TOV%, etc. And this makes sense—you wouldn’t expect any one factor to determine the variance at like .75 or higher in ORTG unless that’s eFG or TS%, which actually have multiple components to them. If you run a test of statistical significance based in a 30 team lesgue, any r of .35 or higher or -.35 or lower will be statistically significant at the usual 5% significance level. And, depending on what we are considering the dataset, the r could be lower and still be statistically significant.

I actually looked at relationship between OREB%, TOV%, FTr, pace, and some other factors and ORTG (I’ll post those some time later) but an r of .4 or even .35 stands out for any season because it doesn’t happen very often. For example, what do you suppose was the highest r between free-throw rate and ORTG was?

.572 in 1998
.524 in 2013
.508 in 1992
.493 in 2018

Or something like pace which an interesting one.

What you're saying makes sense.

I would like to see the correlation between OREB and Net Rtg as well. That will really give us a clue about the net effect of offensive rebounding.

I wanted to keep ORTG and DRTG separate since I wanted to see the effect of offensive rebounding on both offensive rating and defensive rating separately; also, from the graph, whenever the red line exceeds the blue line you know that the net rating will be negative correlation just us when the blue and red lines are far apart you know that the net rating will be more strongly positive. But here you go:

[homecourtloss]

You can also see them both at the same time. You can see in most seasons, higher offensive rebounding % correlated with higher ORTG as one would think but there are seasons when it also correlated with higher DRTG. When the the blue line is inside red line, higher offensive rebounding % correlated more strongly with higher defensive ratings than it did with higher offensive ratings. When you have a big gap, it means that higher offensive rebounding correlated with higher ORTGs and lower DRTGs. Before the 3pt era and from the late 1980s to the latter part of the 1990s, offensive rebounding correlated with higher ORTGs and lower DRTGs. This was also the era when higher ORTGs correlated the strongest with LOWER DRTGs.

That massive gap right around '95 is interesting. Any ideas on why the massive split there?

The only thing I can come up with is that's when they shortened the 3pt line. So suddenly you had teams spamming WAY more three's than they had just a year prior, but they had NOT adjusted their 'crash the glass vs get back defensively' strategy/model to compensate for the change.

The period from the LATE 90's through the mid-late 00's is also interesting for the degree of parity: the gains increased OREB% produces in ORtg seems to be almost precisely mitigated by the diminishing returns in DRtg for about 8-9 years there.

You could be right. There are a couple other things i'm looking at and will post when done.

It's is by far the biggest jump in 3PAr between 1993-1994 and and 1994-1995. The biggest drop at three-point attempt, of course, happen between 1997 and 1998.