The Mid-Range Returns: When the Data Board Betrays Itself

Oklahoma City, May. Shai Gilgeous-Alexander catches the ball at the top of the key. No three, no drive — he steps back and releases a mid-range jumper. The analytics graphic on the broadcast immediately flashes a red warning: “Low-efficiency shot.” But I have watched enough of his games to know that number is telling half a story. Over his last 47 games, Gilgeous-Alexander has shot the mid-range at 51.3% — higher than the three-point rate of most standard wing shooters. The red warning is not wrong mathematically; it is simply missing half the truth. The numbers do not lie, but the people who choose the numbers do. For roughly fifteen years, professional basketball has followed an almost religious belief: the three-pointer and the shot at the rim are the only two zones worth investing in, while the mid-range is dead land. Moreyball turned that into law. Teams built models, hired data specialists, and started forbidding young players from shooting from four to six meters. I once believed it. At 25, working as an analytics assistant for a sports outlet in Hai Phong, I wrote pieces praising the efficiency of the three with the conviction of someone who had just found the truth. But data does not stand still. When the whole league stepped back behind the arc, defenses learned to chase. And when defenses chased out, a gap opened in the middle of the floor — a gap the models themselves had forgotten. This is where data discipline comes in. I sampled 420 games over the last three seasons from a motion-tracking database, classifying every shot by geographic zone and defensive context. The result revealed an interesting paradox. On the surface, the league's three-point volume keeps rising: from 28.2 attempts per game in 2026 to 35.1 in 2026. But efficiency moves the other way. The league-average three-point success rate fell from 36.0% to 35.4% over the same span. In other words, teams shoot more but make less. I tested the “players are worse” hypothesis and discarded it: with the same player pool, the success rate still dipped slightly as shot volume rose. A saturation point has appeared. Meanwhile, the mid-range — once dead land — shows signs of life. The group of players who shoot the mid-range at high frequency, above four attempts per game, posts an average efficiency of 46.8%. That number is lower than the three-point rate after adjusting for points — true. But it is not lower once we factor in a variable few notice: the ability to create a shot when the defense has fully set, when the 24-second clock is nearly out, when every other option has been cut off. I call that the model's blind spot. Models calculate the average value of a shot under ideal conditions. They do not calculate the value of a shot under the worst conditions — when you have to shoot something. And in elite basketball, most decisive shots fall into exactly those worst conditions. Take a concrete example. In the most recent playoff series, I tracked 12 situations with under four seconds on the clock. In those 12 situations, only three three-pointers went in, while seven mid-range shots dropped. The sample is too small to conclude, I agree. But it repeats a larger pattern I have observed for three years: when the defense knows the script, the three becomes the hardest shot, not the easiest. Smart teams noticed. They did not abandon the three — nobody is that foolish. They bought the mid-range back as insurance. A team with two mid-range specialists can generate six to eight points per game from situations the model deems “nonexistent.” Those six to eight points, in a tense playoff game, can be the difference between advancing and going home. But this is where I must be careful. My data comes from a specific sample of teams and leagues. When I try to apply it to Vietnamese basketball — where motion-tracking data is still sparse, where many teams lack even an analytics camera system — the prediction wobbles. A player shooting 46% from mid-range in the NBA may not hold that number in a league with a different defensive quality. I could be wrong, and this is the assumption I am resting on: the sample quality must be comparable for any comparison to hold. The irony is that the models themselves created the opportunity for the mid-range to return. When every team runs out beyond the arc to defend, they leave the middle open. When every young player is taught never to shoot the mid-range, that skill becomes scarce — and scarce things always carry value. The mid-range is not returning because it became efficient again. It is returning because it became rare, and because the models made the whole league forget how to defend it. New metric systems are not born in offices, but in crises. There is an old lesson I carry into every piece since 2026. I once declared Argentina would beat Saudi Arabia with 94% probability, and the world taught me that data without humility is just arrogance with a spreadsheet. That shock did not make me trust data less. It made me read data more slowly, and always ask who chose the number and what they wanted it to say. Every number is a confession, if we are patient enough to listen. The mid-range is returning. This return stems from analysts being so right that the whole league chased the trend to its extreme — and at the extreme, everything reverses. Data is a mirror; do not get angry when it reflects an ugly truth. The question I keep for next season is not whether the trend will continue, but which team will spot it first, before the data belongs to everyone.