The mid-range jumper isn't dead. It just got expensive.
The standard narrative about the modern NBA is that analytics killed the mid-range jumper. The math said two-pointers from outside the paint were the worst shot in basketball — a slightly higher conversion rate than threes, but multiplied by a value of two instead of three, which makes their expected points per attempt lower than either a layup or a corner three. So coaches stopped designing plays for them, and players stopped taking them, and the mid-range disappeared.
That story is half right. The mid-range did collapse — league-wide attempts from sixteen feet to the arc fell by more than half between 2010 and 2020 — but it didn't disappear. It got concentrated. The players who still take mid-rangers in high volume are mostly the league's best scorers, and the reason they take them is not that they ignore the math. It's that they understand a part of the math that the league-average version misses.
What the simple model says
The expected-points argument is genuinely robust. A league-average mid-range jumper goes in roughly 40% of the time, producing 0.8 points per attempt. A league-average corner three goes in roughly 38% of the time but pays three points, producing about 1.14 points per attempt. A league-average shot at the rim goes in north of 65%, producing well over 1.3 points per attempt. Two of those three numbers are much bigger than the third. If you treat every player as league-average, the mid-range is obviously the wrong shot.
The argument led to a specific kind of offensive philosophy in the 2010s — drive and kick, hunt corner threes, get to the rim or get to the line, never settle. Teams that ran it well won championships. Teams that didn't lost a lot of regular-season games against teams that did. The mid-range was, for several years, the textbook example of an inefficient shot that the analytics had finally banished.
What the simple model misses
Two things, both about what happens late in possessions. The first is that mid-range shots become available in situations where corner threes and rim attempts don't. Late in the shot clock, against a defense that has loaded up the paint and chased shooters off the line, the mid-range pull-up is often the only shot a possession can generate without turning the ball over. The expected value of "the only shot you can get" is the comparison the simple model misses. It isn't 0.8 points versus 1.14. It's 0.8 points versus whatever a broken late-clock possession produces, which is much less than 1.14.
The second is that the simple model assumes a league-average shooter taking the shot. That assumption is fine for league-wide policy questions and exactly wrong for individual decisions. An elite pull-up shooter doesn't convert at 40% from the mid-range; they convert at 48% or higher. At 48%, the math reverses. A 48% mid-range is worth 0.96 points per attempt, which is more than a league-average corner three from a non-shooter forced to take it under pressure. The same is true at the rim — an elite finisher gets there at a higher conversion rate, but they also draw more contact and more help, which depresses the rate further. The gap narrows.
The Durant and Mitchell archetype
Watch a possession involving Kevin Durant in any season of the last decade and you'll see the mid-range argument refute itself in real time. The defense knows he's going to pull up from fifteen to twenty feet. They sell out to take it away. He gets it anyway, because his release point is over almost every defender and his shot doesn't require space. He converts at a rate that makes the shot worth more, per attempt, than the corner threes the rest of the league is chasing.
Donovan Mitchell is the next-generation version. So is DeMar DeRozan, late-career. So is, in flashes, every elite scoring guard in the league. The skill these players share isn't a refusal to read analytics. It's the recognition that the analytical floor is set by league-average shooters and the elite-shooter ceiling is somewhere the math actually rewards. The Spurs front offices and the Heat front offices know this. They're not telling DeRozan to take fewer mid- rangers. They're telling the players around him not to.
The team-level math
Where this gets interesting is that the team-level math is different from the player-level math. A team that takes a lot of mid-rangers is mostly a team whose mid-rangers are not being taken by elite pull-up shooters, because there are only ten or fifteen of those in the league at any given time. The mid-range still loses, on average, at the team level. It wins, on average, for the small subset of players who can take it efficiently. The right team policy is "ban the mid-range for most players, license it for two." That's approximately what every modern coaching staff actually implements, whether or not they describe it in those terms.
The playoff exception
Playoff defenses change the math again. In the regular season, defenses can't fully sell out to take away corner threes and the rim; they don't know the personnel that well, they're playing on short rest, and there's a new opponent every other night. In the playoffs, defenses spend seven games studying a single opponent and they can erase specific shots from a specific player's diet. Corner threes get harder to manufacture. Rim attempts get walled off. The possessions that used to end in good analytical shots end in worse analytical shots — including, sometimes, mid-rangers. Teams whose stars can't make those shots cap out at the second round. Teams whose stars can keep playing.
This is the second-order argument the early-analytics era didn't fully see: that an offensive system optimized for the regular-season shot diet might be exactly wrong for the playoff shot diet, and that the value of mid-range shooting is partly insurance against playoff defenses. The clearest evidence is that the teams with deep playoff runs in the last several years almost all have at least one elite mid-range shot creator. The teams that don't tend to bow out when a defense locks in.
What the math probably says next
Two predictions. The first is that the mid-range comeback we've already seen — league-wide attempts have ticked up slightly since bottoming out — will not turn into a full reversion. The simple math still wins for league-average shooters, and the league-average shooter still takes most of the league's possessions. The second is that the next generation of analytical insight here will be more granular: not "is the mid-range good or bad" but "is this mid-range, at this point in the clock, for this player, against this defender, good or bad." Tracking-data shot-quality models already do a version of this. The next decade of front-office work will be making them precise enough to inform play design in real time.
The slogan "the mid-range is dead" was always wrong as a descriptive claim. It was right as a normative one for most players. That distinction is the part of the analytics revolution that broadcasts still don't quite communicate. The math didn't say nobody should ever shoot a mid-range. It said most people shouldn't, and a few people should, and figuring out which is which is the actual job. The league has spent fifteen years working it out, and the answer turns out to be more interesting than the slogan.