Nylon Calculus: More lessons from NBA rebound tracking

Analyzing a new cache of NBA rebound tracking data can help identify optimal strategies for shot selection and positioning rebounders.

It’s been said that the NBA is a ‘make or miss’ league. And that’s true — shot-making is a crucial factor in determining the outcome of every game. But it could also be said that it’s a ‘make or miss…and then rebound or not’ league (although, admittedly, that would be kind of a mouthful).

Rebounding is useful. The league-average possession has generated 1.10 points this season, whereas possessions extended by an offensive rebound have yielded more — 1.19 points, on average (note: on the flip side, possessions ending in a defensive rebound have produced a total of 0 points so far this year).

But rebounds are unpredictable. A missed shot caroms off the rim or backboard and ricochets to an arbitrary spot on the floor. Where will the ball end up? Nobody knows.

Or do we? I mean, rebounds aren’t totally random, right? Surely there are some patterns to be observed — connections to be made between the location of a missed shot and the most likely landing place of the resulting rebound. With the right shot-tracking data, we might even be able to create strategies that optimize a team’s chance of capturing an offensive rebound.

Luckily for us, Darryl Blackport (@bballport) — the architect responsible for the vast warehouse of facts and figures over at PBPstats.com — has begun an ambitious rebound-tracking project that might help us do just that. So far, he’s logged the coordinates of nearly 4,000 missed shots along with the coordinates of all the resulting rebounds…BY HAND. To get started, he’s focusing his attention on any games played by the Raptors and Lakers this season.

So, what can this unique data resource tell us about rebounding?

Well, let’s start with what we already know: long shots lead to long rebounds. Kirk Goldsberry confirmed this old adage a few years ago using SportsVU rebounding data from the 2013-14 season. He showed that longer shots do indeed yield longer rebounds. We see the same trend in the hand-tracked rebounding data Blackport has collected this season.

Interestingly, there is also a relationship between shot distance and offensive rebounding rate, something which was previously demonstrated by Rajiv Maheswaran, the CEO of Second Spectrum, in a 2012 Sloan paper. Check out how shot distance has been linked to the likelihood of an offensive rebound this year:

The best opportunity to grab an offensive rebound is after a short shot — as 35 percent of the misses from inside of four feet were secured by the offense. In contrast, mid-range misses from 13 to 21 feet were recovered by the offense half as frequently, only about 17 percent of the time. On shots from any further away than that, the likelihood of an offensive rebound increased again, inching back up to about 25 percent on 3-pointers from 28 feet or more. This is consistent with what Blackport, Mike Beuoy, and Tom Haberstroh found in their previous research on ultra-long-distance 3s. They used play-by-play data to show that these so-called “buy one, get one” 3-pointers were more likely to create offensive rebounds than misses from shorter distances.

Here’s the neat thing: We can use Blackport’s new, hand-tracked rebounding data to help untangle the relationships between shot distance, rebound distance, and the probability of an offensive rebound.

The chart below summarizes the results of 1783 missed 3-pointers in the rebound tracking database — how far did the rebounds travel and which team grabbed them once they got there?

On shots from 22 to 39 feet, the most likely rebound distance was seven to 21 feet away from the basket (41 percent of the long-range shots were recovered in this area). Defensive players tended to own this rebounding battleground just outside of the restricted area. If the ball was recovered at this distance, it was usually because the defender was boxing out or at least in good position relative to the potential offensive rebounders. As a result, of the 733 missed 3-pointers that were captured seven to 21 feet from the hoop, only 22 percent went to the offense.

If the rebound was made closer to the basket (as occurred on 310 occasions, 17 percent of the time), the offense had a better chance of grabbing the ball (21 ORB%). This included some instances where the defense was unable to maintain good position, resulting in tipped shots and follow-up dunk attempts.

Although it was rare for a rebound to clear the 3-point line (it happened on just 58 missed 3-point attempts, 3 percent of the time), the offense was at a distinct advantage in these situations (69 ORB%). This makes sense intuitively, because the offense will tend to have more players standing beyond the 3-point line than the defense and this would have been especially true just after a long shot was attempted.

The next chart summarizes the results of the 779 missed midrange shots.

For the midrange shots, we see many of the same patterns that we saw for the 3-pointers. The zone from four to six feet was the most likely rebound destination (40 percent of the mid-range misses) and the defensive rebounders dominated that area of the court on these slightly shorter shots as well (12 ORB%). As above, if the rebound went beyond 6 feet, the offense had better luck (23 ORB% on 181 rebounds from seven to 21 feet; 67 ORB% on 15 rebounds from 22 to 39 feet). Compared to the missed 3-pointers, a larger proportion of the missed mid-range shots (35 percent) were corralled inside the restricted area and the offense had marginally more success in there (25 ORB%).

The third chart summarizes the results of the 249 shots missed from four to six feet.

On missed shots from inside the paint, the most likely landing zone for rebounds was directly adjacent to the basket (132 instances from zero to three feet, 53 percent of the time). With the shooter having an opportunity to crash in and gather his own board on these shorter shots, the success rate for the offense rebounders climbed (38 ORB%). Defenders continued to be dominant in the four to six-foot rebounding area, but the offensive rebounding rate was slightly better on these paint shots (19 ORB%) than it was on the missed midrange (12%) and 3-point jumpers (14%) which we examined above. It was rare for the paint shots to rebound past six feet (only 15 percent of the misses traveled seven to 21 feet) and the offense struggled to collect them when they did (11 ORB%). Some of these long rebounds must have bounced past the shooter to be scooped up before reaching his teammates outside the 3-point line.

The last chart summarizes the results of the 883 shots missed near the rim.

Missed shots that came from inside the restricted area tended to be rebounded before they left the restricted area (53 percent of the time). In these instances — when the shooter was already well-positioned to tip in or dunk his own miss — the offensive team won the rebound nearly half of the time (47 percent). The ball traveled surprisingly far on a number of these shots missed at the rim (landing from four to six feet 26 percent of the time and from seven to 21 feet 19 percent of the time), presumably as a result of spectacularly botched dunks and alley-oops. When the ball bounced beyond the reach of the shooter like this, the defense tended to control the rebound (18 ORB% from four to six feet, 22 percent from seven to 21 feet) unless it continued all the way out to his teammates at the 3-point line (56 ORB% past 22 feet).

Based on this analysis, what recommendations might we give a team who wants to win more offensive rebounds? Here are three ideas:

• Stop shooting long mid-range shots, because they are less likely than 3-pointers to end up as an offensive rebound.
• Drive to the basket more often to create more offensive rebounds.
• Shoot long 3-pointers and then send offensive rebounders to ‘soft crash’ the area around the free-throw line to collect any long rebounds that may result.

The first suggestion — to shoot fewer mid-range shots — was proposed by Maheswaran and his team in their Sloan paper on rebound tracking. It’s probably a valid assertion technically, but it feels a bit like beating a dead horse, doesn’t it? We get it, no mid-range shots allowed! And anyways, the offensive rebounding rate wasn’t really that much different when the shooter moved from the long mid-range back to 3-point land.

The second suggestion — to drive to the rim as frequently as possible — also makes some sense. Attacking the rim forces help defenders to rotate, leaving offensive rebounders free to hit the glass. Plus, if a player shoots the ball near the rim, he’ll be right there to try to follow any of his misses. The complication is that shots which are missed or blocked at the rim come at a cost — they tend to result in more points for the opponent on the next possession. Balancing this tradeoff would definitely require careful judgement.

The last suggestion — the soft crash — might be the best advice of the three. Shots missed inside the arc end up as long rebounds (in the area from seven to 21 feet) about 20 percent of the time, but shots missed outside the arc become long rebounds twice as frequently, 41 percent of the time. So, when a 3-point shot goes up, why not send an extra rebounder into that middle area? This would increase a team’s chance at securing offensive rebounds, without necessarily sacrificing their ability to have stout transition defense. All the reward with none of the risk.

For more basketball stories told through analytic insights, check out Nylon Calculus and follow us on Twitter and Facebook.