← All articles

Opponent three-point percentage is mostly luck

Every season produces a handful of teams celebrated for their perimeter defense, and the evidence offered is almost always the same: opponents are shooting a low percentage from three against them. The narrative writes itself. The wings are long, the rotations are crisp, the coach has built a culture of closing out. Then, a few months later or the following October, the number drifts back toward the league average and the story quietly changes. The roster did not get worse at defending. The number was never mostly about defending in the first place.

Who actually takes the shot

Start with the simplest point. A three-point attempt is released by an offensive player, and whether it goes in depends overwhelmingly on that player’s skill, his mechanics on that particular night, and the distance and angle of the shot. The defender can influence the attempt, but once the ball leaves the shooter’s hand there is nothing left to defend. Even an elite shooter misses well over half of his threes, and an average one misses closer to two of every three. That is an enormous amount of built-in randomness on every single possession, and it belongs to the offense.

Now aggregate it. A team faces somewhere in the range of thirty-plus opponent threes a night in the modern league, which sounds like a lot until you treat each attempt as a weighted coin flip. Over a month, the difference between opponents shooting a few points above or below their true talent can be produced entirely by chance, with no change in defensive quality at all. Over a full season the sample is larger, but the spread of outcomes that pure variance alone would generate is still wide enough to swallow most of the differences between teams.

Stability is the tell

The cleanest way to ask whether a statistic measures a skill is to split the sample and see whether a team’s number in one half predicts its number in the other. Skills persist; noise does not. When analysts run this exercise on defensive three-point percentage, the relationship between the first and second halves of a season tends to be weak, and the relationship from one season to the next is weaker still. In most recent seasons, a team’s opponent three-point percentage in November has told you surprisingly little about what it will be in March.

Compare that to the stats a defense produces by its choices rather than by the outcome of the opponent’s shot. How many threes a team allows per possession, what share of opponent shots come from beyond the arc, and how many of those come from the corners are all far more consistent within a season and across seasons. That contrast is the whole argument in miniature. The parts of perimeter defense that repeat are the parts the defense decides; the part that bounces around is the part the shooter decides.

What the percentage is secretly measuring

Opponent three-point percentage is not pure noise, but the signal it does carry is mostly not the signal people think. Three things beyond defensive effort shape it heavily. The first is schedule. A team that happens to face a run of poor shooting rosters, or catches good shooting teams on the second night of a back-to-back, will post a flattering number that says nothing about its scheme. Over a full season schedules even out somewhat, but over the twenty-game stretches where most narratives form, they do not.

The second is shooter quality within the opponent. Defenses that funnel threes toward a team’s weakest shooters are doing something genuinely smart, and it shows up as a lower allowed percentage. But that effect is much smaller than it looks, because coaches cannot fully control who ends up with the ball in rotation, and good offenses actively hunt their best shooters’ looks. The third is shot location, which matters enough to deserve its own section.

Corners, above the break, and openness

Not all threes are the same shot. The corner three is shorter than the above-the-break three, and across the league it goes in at a noticeably higher rate. It is also usually the product of defensive breakdown: a corner shooter gets open when his defender helps on a drive or a roll and the rotation behind him is late. A defense that prevents corner threes is showing real discipline, and that discipline tends to persist from season to season because it reflects scheme and personnel rather than chance.

Openness works the same way. Tracking data that classifies shots by the distance of the nearest defender consistently shows that wide-open threes go in more often than tightly contested ones, and that defenses differ in how many open looks they surrender. That is the lever a defense actually controls. The catch is that the shooting percentage on those open looks is still up to the shooter. A team can force contested threes all season and watch opponents bank them in, or concede wide-open looks and watch them clang. The process is good or bad independently of the result on any given night, and the process is what will carry forward.

Put differently: a defense should be judged on the quality of the shots it allows, not on whether those shots happened to fall. If you know how many threes a team gives up, how many of them come from the corners, and how open they tend to be, you can estimate an expected opponent percentage. The gap between that expectation and the actual figure is mostly luck, and luck is the part that regresses.

Why the narrative keeps getting built

The appeal of opponent three-point percentage is that it is visible, easy to rank, and attached to a story everyone wants to tell about effort and length. It also correlates with winning in the short run, because a stretch of cold opponent shooting really does produce wins, and wins invite explanation. Nobody writes a feature about a team that went 12–3 because opponents missed open threes. They write about the defensive identity, and the identity becomes the explanation for a number that was largely variance.

The cost of that mistake is real. Teams that believe a lucky defensive stretch are prone to overpaying defenders whose reputations were inflated by it, or to sticking with a scheme that concedes too many threes because the shots happened not to fall. The reverse happens too: a sound defense that suffers a hot-shooting month gets labeled soft, and its coach starts changing coverages that were working. Both errors come from treating an outcome as a measurement.

How to read it

When you size up a team defense, put more weight on volume than on accuracy. Check how many three-point attempts a team allows relative to the league and how that figure has held up over the season, and treat the allowed percentage as a weather report rather than a scouting report. If a team’s overall defensive rating looks elite while its opponents are attempting plenty of threes at a suspiciously low clip, expect that rating to drift. If a team is allowing few attempts but opponents are converting them at a high rate, the defense is probably better than it looks.

None of this means perimeter defense is fake or that closeouts do not matter. Contesting shots works, and over very large samples a defense’s ability to make threes harder does leave a mark on the percentage. The point is that the mark is faint relative to the noise around it, and that the cleaner evidence of good perimeter defense lives one step earlier in the possession: in the threes that never got taken, the corner looks that never opened up, and the shooters who found a hand in their face. Judge a defense on those, and let the shooting percentage settle wherever chance decides.