A theme I’ve already come back to repeatedly this season is that measuring defense is hard. In most cases, we’re trying to identify what the offense isn’t doing and then determine how much of the credit is due to defenders individually and collectively and how much is normal variation in offensive play, which from the standpoint of the defense can essentially be termed “luck.”
Advertisement
This exercise is hard enough when offensive strategies are static — if we know what the offense is attempting to do, variations in, for example, their shooting profiles and accuracy can be compared to a stable baseline. In that case, the general parameters of #MakeOrMissLeague randomness are well known.
The task becomes far more difficult when offensive strategy changes. In the evolving NBA, this is where we find ourselves. Previously, defensive analysis was merely trying to suss out the impact of gusty winds on an archer hitting a target. Now, it seems the target itself might be moving.
Even as the 3-point revolution gathered steam, there seemed to be some fundamental maxims that could help evaluate defense. At a high level of abstraction the theory goes something like this:
- By and large, NBA players make reasonable decisions about when to shoot.
- Given the well-established incentives at work — scoring points still equals earning dollars, even if the other considerations are becoming more determinative of salary as we better understand efficiency and the marginal value of possessions used — players will shoot shots they think they can make. (Speaking of incentives, shot clock turnovers are not charged to players, so stop dodging the ball at the end of blowouts, please and thank you.)
- Part of the differentiation between NBA and near-NBA level players is the ability to correctly make these sorts of decisions more often. Players whose appetites for scoring that far outstrip their abilities tend to not be in the league, at least not for long. You have to be on the floor to contribute and taking hopeless shots is a good way to end up sitting next to an assistant coach on the sidelines.
- Because NBA-level players take shots they more or less correctly feel they can make, it stands to reason that the shots they do not take are declined because something about the situation is off. Maybe a pass pulls a player off balance, or he doesn’t catch the ball cleanly so that he isn’t in rhythm. Perhaps there are time and score aspects that cause him to defer. And of course, maybe the defense is playing him too tightly for the player to feel sufficiently confident about making the shot to attempt it in the first place.
- Fundamentally, this is why measuring a team’s defense by opponents’ accuracy on jump shots is flawed, as the things which will tend to depress accuracy will also induce them to do something else with the ball other than just catch and shoot.
- Thus, defensive shot profile is often a better indicator of defensive integrity than the percentage allowed. The fact that defensive shot profile stabilizes much more quickly than accuracy allowed certainly adds empirical support to this line of thinking.
Over time, this has served as a perfectly cromulent theory of NBA defense, at least on the perimeter. For a number of reasons, shots at the rim don’t create the same incentives between higher foul draw rates and the fact that even heavily contested shots near the rim are scored at reasonable accuracy:
This is why actively deterring shots at the rim once a player gets there with the ball is a relatively rare skill, while basically anybody can “run a guy off the line” to prevent a 3-pointer.
Or rather, anybody could run players off the line. In the course of discussing early season defensive shot profiles, I noted the tendency of games against the Rockets to confound normal analysis of defensive shot profiles, because no team in the NBA exemplifies the “let it fly” ethos to quite the degree Houston has over the years. To run a particularly telling chart back, in recent years, Houston has lapped the field in terms of putting up contested 3s:
In the early days of this season, not much has changed, Houston still leads the league by a wide margin in overall contested 3s. However, there does seem to have been a subtle shift in the league as a whole:
Though both open and contested 3s have risen as a proportion of overall shots over time, so far this season, we have seen a somewhat sizable increase in contested 3-point attempts while for the first time there has been a decline in uncontested attempts, at least as a proportion of shots.
So what gives? Given the continued rise in 3-point attempts (entering Wednesday’s games, 37.3% of all shots had been 3s, up from the all-time record of 35.8% in 2018-19), increased aggressiveness on the part of shooters almost has to be a major component, though the degree to which defenses are more effectively closing out to shooters is less clear. After all, to measure the defensive aspect, we’d have to know something about the potential shots which aren’t being taken. If perimeter touches are being more heavily pressured, then maybe the defense is driving a large share of this effect, whereas if perimeter catches are largely the same, the increase in contested 3s is almost completely about offensive decision-making.
Advertisement
Unfortunately, public data doesn’t allow for direct analysis of this phenomenon, and I can tell you from experience that even with the full tracking set, defining a “shooting opportunity” is quite tricky. However, there are some indications suggestive that the change is more and more offenses working with a hair trigger. For starters, the chart below summarizes the team-by-team increase in contested 3s as a proportion of all shots:
Four teams (Dallas, Washington, Minnesota and New Orleans) have all upped their contested-3 attempts rate by more than 5 percentage points. To reframe it, that’s more than one additional contested 3 per quarter. It’s a big change. If the league-wide change was tighter attention being paid to potential shooters, I would expect the above chart to be somewhat flatter. While there would be some team-by-team variation driven by changes in personnel, the spread between teams would be narrower.
More interesting is looking at shooting propensities early in the shot clock. While with the clock running down we could more easily ascribe a rise in contested shots to more effective closeouts as offensive players dealt with teammates dealing them “live grenades” at the end of the 24. With the element of shooter choice removed, or at least reduced by the need to get an attempt up, that’s a plausible story. However, late clock contested 3s have held steady at around 4% of all shot attempts since 2013-14. However, contested 3s earlier in the shot clock have risen sharply:
Earlier in the shot clock, while the offense still has options, is where the entirety of the increase in contested 3-point attempts is occurring (save for a minor blip in early clock attempts, which is probably attributable to the change in shot clock reset from 24 to 14 after an offensive rebound prior to 2018-19). Considering that contested 3-point attempts represent roughly league-average efficiency shots over the last several years, the rise of early clock 3s could very much be described as teams and players “settling” for OK shots. Returning to an earlier chart:
This is turning the mantra of “turning down a good shot to get a great shot” on its head. An increasing number of teams appear to perfectly willing to “let it fly” at the earliest reasonable shot:
To bring the discussion of full circle, how do we judge defensive performance against this changing offensive profile? Should it be considered a good or bad defensive possession to give up a roughly average efficiency shot early in the shot clock? I wish I had a good answer, but I don’t yet. As is typical with measuring defense, the more we think we learn, the more we realize we don’t yet know.
(Photo: Jane Tyska / Digital First Media / The Mercury News via Getty Images)