Three-Point Defense: Can You Actually Defend the Three?

Watch enough postgame interviews and you'll hear a coach take credit for "running them off the line" after a win, or a fanbase rage about a defense that "let them get hot" after a loss. Both are reaching for the same idea: that a defense controls whether the other team makes its threes. The data says one of those statements is mostly true and the other is mostly an illusion. A defense can powerfully shape how many threes the opponent takes and where from. Whether those threes go in is, to a large and uncomfortable degree, out of its hands.

The finding: attempts are controllable, percentage mostly isn't

This is one of the more durable results in basketball analytics, and it surprises people every time. A defense's opponent three-point rate — the share of opponent shots that come from behind the arc, and the location of those shots — is a relatively stable, repeatable trait. A team that suppresses three-point volume in one stretch tends to keep doing it. But a defense's opponent three-point percentage — whether those threes actually go in — is far noisier and regresses heavily toward league average over time. Year to year, a team's opponent 3P% is much less repeatable than its opponent rim rate or the three-rate it allows.

The reason is structural, and it ties directly to our piece on three-point variance. A three is a high-variance event by nature; once it leaves a competent shooter's hands, the outcome bounces around hard. A defense can affect the inputs to that shot — how open it is, how rushed — but it cannot reach out and adjust the flight of the ball. So the controllable part of three-point defense lives almost entirely before the shot goes up, in the decision of whether the shot happens at all and what kind of look it is.

What a defense can actually control

If percentage is mostly noise, what's the job? Quite a lot, as it turns out — just all of it upstream of the shot:

• Three-point rate allowed. The single most controllable lever. Force ball-handlers into the mid-range and the paint, take away the kick-out, and you simply reduce how many of the most valuable shots the opponent gets up. Fewer attempts at a noisy-but-high-value shot is a real, repeatable edge.
• Contest rate and closeouts. A defense can systematically get a hand up. The gap between a wide-open three and a contested one is meaningful in the aggregate, and forcing the contested version is a process the defense owns even if any single contested shot might still drop.
• Limiting corner threes. The corner three is the shortest, most efficient three on the floor. A defense that refuses to surrender corner looks — through disciplined help rotations and "no-middle" or "no-corner" schemes — is steering the opponent toward lower-value above-the-break attempts. That's shot-quality control, not luck.
• Suppressing wide-open looks. Wide-open threes convert at a markedly higher clip than guarded ones. A defense can't change the make rate on a given openness level much, but it can change the distribution of openness — how many of the opponent's threes are wide open versus contested. That distribution is a process the defense influences.
• Forcing teams off the line and into twos. The flip side of all of the above: a great perimeter defense converts would-be threes into long twos and drives into traffic, the least efficient shots in the game.

Notice the pattern. Every item is about the shot that gets taken. None of them is "make the open shooter miss," because once you've allowed an open shot, the result largely belongs to the shooter and the variance gods.

The arithmetic of why percentage fools us

The following uses clearly-labeled round numbers to illustrate the mechanism — these are not real team figures. Imagine two defenses over a stretch of games. Defense A allows opponents to attempt 30 threes a night; Defense B allows 40. Suppose, for the sake of the illustration, that opponents shoot a league-ish 36% against both, because neither can truly control the make rate. Defense A surrenders about 10.8 makes (32.4 points); Defense B surrenders about 14.4 makes (43.2 points). That's a roughly 11-point gap from beyond the arc — created entirely by the attempts allowed, with the percentage held identical.

Now run it the other way. Hold attempts equal and let the opponent percentage swing from a cold 32% to a hot 40% — a swing that, as the variance piece explains, happens constantly for reasons that have little to do with the defense. On 35 attempts that's the difference between about 11.2 and 14 makes, another double-digit point swing, except this one the defense mostly didn't earn or cause. The lesson: when you see a defense's points-allowed-from-three move, ask whether it moved because of attempts (a repeatable, controllable signal) or percentage (mostly transient noise). They look identical on the scoreboard and mean opposite things.

How this should change the way you watch

The single-game implication is the big one. When a team loses because the opponent "couldn't miss from three," the most likely explanation is not that the defense suddenly forgot how to close out — it's that a high-variance shot ran hot for one night, exactly the kind of small-sample noise our three-point variance piece is built around. The same is true in reverse: a defense that "locked them up from deep" for a game may simply have caught a cold shooting night. A series is a few more flips of the same coin, not enough to make the noise disappear. Before crediting or blaming the defense, the honest question is whether the opponent's shot quality — openness, location, shot type — was actually different, or whether only the makes were.

For evaluating a defense over a season, the move is to weight the repeatable inputs over the noisy output. A unit that consistently holds down opponent three-point rate, forces contests, and refuses corner looks has a real, sticky skill — even across a span where its opponent 3P% happened to run high. Conversely, a defense whose gaudy ranking rests on a suppressed opponent 3P% is a regression candidate, because that number tends to drift back toward league average no matter how good the defense looks today. Reading the shot diet a defense forces — how much of the opponent's offense it pushes off the line and into low-value twos — tells you far more about the unit than the percentage those threes happened to fall at.

The takeaway

You can absolutely defend the three — by defending the attempt. Run shooters off the line, take away the corners, contest what's left, and herd the offense into the shots no one wants. What you mostly cannot do is reach into a clean release and make it clang; opponent three-point percentage is largely the shooter and the bounce, and it regresses. So credit defenses for the looks they prevent and the looks they degrade, stay skeptical of a hot or cold three-point night as evidence of anything, and remember that the durable part of three-point defense happens entirely before the ball is in the air.