Deep Dives: Floor Balance and Transition Defense

The transition from offense to defense has long been one of the topics in NBA stats which fascinates me the most. While possession-based analysis has be the key to unlocking many mysteries of the game, the extreme focus on a discrete my turn-your turn cataloging of events has meant the space between has been left behind.  Of course we know how a possession starts can have a sizable impact on how it might end – playing against a set half-court defense is demonstrably more difficult than playing in the open floor after a steal, to the tune of around 30 points per hundred possessions, greater than the distance between the best and worst teams.

This insight naturally applies to both sides of the ball – steals are good because they lead to offense, while forcing deadball turnovers is never bad thing, it’s just not nearly as good. Meanwhile, on the offensive end, finishing possessions with bad floor balance can compromise one’s own defense. These insights were fairly well-understood before they were formally quantified, but the analysis has never really gone much deeper.

A topic I’ve returned to from time to time is the offensive rebounding-defensive transition tradeoff, but that has always been done in the realm of aggregates, looking at the totality of a team’s possessions. During last night’s games, Adam Mares remarked that it seemed to him that the Nuggets were particularly susceptible to transition after midrange attempts:

He asked me if there was a way to determine if his intuition in this regard was true. This is a fabulous question as it starts to get to the heart of the matter in terms of determining if teams are particularly vulnerable in certain circumstances. Again, we know that certain actions in general tend to lead to opposing fast breaks, such as a point guard missing a shot at the rim, but are some teams specifically susceptible even given the broader trends? And moving back to the more general, just how much of the “good transition defense” exhibited by some teams is merely a residual of their shot selection?

To investigate, I looked at all shots which met the following conditions

1. The previous play was a live ball defensive rebound of a missed field goal attempt, as dead balls, including team rebounds, result in a very different mix of ensuing possessions.
2. The shot itself was the next event in the official play-by-play log.

For the preceding shot location, I divided the court into areas – Above the Break threes, Top of the Key[2. midrange shots above the free throw line.], Baseline[3. All shots outside of the paint from below the FT line. Initially I had this separated into corner 3s and baseline midrange shots, but the averages of early shots following misses from inside and outside the arc on the baseline is virtually identical near 24%, so I collapsed them into one area given the relatively small number of corner threes.], shots in the Restricted Area, and all other paint shots. Further, I defined “early” shot attempts as those coming in seven seconds or less from the rebound.[4. Across multiple studies, I’ve found that after the first seven seconds or so, the observed shooting efficiency of possessions doesn’t much depend on the prior play. Moreover, the offensive benefits from steals and defensive rebounds accrue as a result of both more “early shots” and greater success on those shots.]

Initially, I was going to look at both the proportion and efficiency of early shots allowed by preceding shot location, but quickly decided to focus on just the proportion. The team-by-team samples were so small in each bucket that the statistical noise overwhelmed the signal. However, on aggregate, shots from defensive rebounds after longer than seven seconds showed an eFG% of 49.5%, while “early” shots had an eFG% of 57.5%, which is entirely consistent with past findings.

In total, the leaguewide percentages shake out as follows:

To some degree the “miss a layup at one end, give up a dunk at the other” adage seems to have some truth. But these are just general percentages, are there individual teams that do better or worse? Does Denver, for example, give up more than their share of transition play after missed midrange shots? In fact the Nuggets do, though not by much. So far this season, 22.4% of opponent shots following Denvey midrange misses have come early, nearly 2% higher than normal, ranking 24th in the league. However, the bottom four of Brooklyn (25.3%), Philadelphia (25.8%), Phoenix (27.4%) and the Lakers (27.8%) are separated by some distance from the rest of the league. In fact, Denver has only allowed about 5 more “extra” transition plays over the course of the season from these top of the floor midrange shots than if they had been average at getting back.

In fact, the Nuggets have been slightly better than average at recovering following their own misses, giving up early shots on 26.7% of chances as compared to league average of 27.5%. Adjusting for shot location[5. Using existing rates from each zone, but assuming a league average distribution of attempts between zones to not overly penalize teams for getting a high proportion of shots right at the rim.], Denver jumps to ninth! Conversely, the Clippers are 20th in terms of raw percentage, but accounting for their tendencies towards shooting from the top of the key rather than getting into the paint, they fall to 26th. The following chart shows the performance of each team in terms of allowing transition shots by preceding shot zone:

In terms of overall magnitude, the difference between the best and worst, it works out to around three extra transition opportunities per game, or about .5 points. Overall not a huge gap, but the contours of each teams’ performance do show some potential areas of vulnerability.

For example, the Cavaliers are about average overall, allowing transition chances on 27.9% of opponents’ defensive rebounds. However, looking more closely, there is a dramatic split by shot location, with the team ranking among the ten best in terms of getting back and defending on shots taken above the free throw line, but below the average on baseline and paint shots, including allowing 50.6% transition% on missed shots in the restricted area, higher than any other team including the hapless Lakers and Sixers. In a closely fought playoff series, such as against, who knows, the Warriors, this proclivity for giving up fast breaks on missed shots at the rim could easily prove to be a problem.

Briefly before wrapping up, the reverse side of this equation is similarly fascinating in terms of looking at where teams are able to runout on offense after forcing a miss. The chart below shows teams’ percentages following a defensive rebound of opponent misses from each area:

 
As should be obvious, the spread between teams is much wider on offense than defense. In particular, Utah stands out. The Jazz play slow from any situation it seems, but their inability to score quickly after forcing misses at the rim – something a team featuring Rudy Gobert and Derrick Favors will always be quite good at – prevents them from maximizing the effect of their great rim protection. Whether this is the sort of stat which might improve as Utah upgrades at the lead guard spot, or is simply a result of the sort of deliberate style of play imposed by playing two traditional bigs remains to be seen, but it is a cause for concern.