Every revolution in sport eventually gets reduced to a chart. Basketball’s is a shot map: two dense clusters, one at the rim and one beyond the three-point arc, with a conspicuous desert in between where the mid-range jumper used to live. That map is the visible output of the most successful analytics movement in team sport, and the story of how it happened is really the story of a simple piece of arithmetic taking four decades to conquer a game.
The arithmetic is this: a three-point shot is worth 50 per cent more than a two. A league-average three-pointer made at around 36 per cent yields roughly 1.08 points per attempt; a long two-point jumper made at 40 per cent yields just 0.80. Once you frame every shot as an expected-value proposition, the mid-range jumper, the signature shot of basketball’s golden eras, becomes the least efficient option on the floor, behind only contested heaves. The best shots, mathematically, are layups, free throws and threes. Everything that has happened to basketball’s shot diet since flows from that observation.
From novelty to default
The league did not see it immediately. When the NBA introduced the three-point line in 1979-80, teams attempted just 2.8 threes per game, and the shot was treated as a gimmick for decades. Adoption crept upward through the 1990s and 2000s, but the curve bent sharply in the 2010s: by 2018-19, teams were attempting 32.0 threes per game, more than ten times the original rate, and by the 2024-25 season roughly 42 per cent of all NBA field-goal attempts came from beyond the arc. Interestingly, the curve briefly flattened, with 2022-23 marking the first year-on-year decline in three-point attempts in more than a decade, before the upward trend resumed, driven increasingly by pull-up threes replacing pull-up twos.
Moreyball: the experiment that proved the maths
The acceleration has a name attached to it: Daryl Morey. As general manager of the Houston Rockets, Morey built rosters and offences explicitly around expected value, a philosophy inevitably dubbed “Moreyball” in homage to baseball’s Moneyball. The commitment was total. Beginning in 2013-14, Houston led the NBA in three-point attempts and makes in six of the next seven seasons, while ranking dead last in mid-range attempts in each of those seven years. The Rockets never won a title, but they won enormously and repeatedly, and the rest of the league absorbed the lesson. Within a few years, what had been a Houston eccentricity was simply how basketball was played; the Golden State Warriors, with Stephen Curry weaponising the pull-up three, supplied the championships that made the style aspirational as well as rational.
The infrastructure behind the shift matters as much as the philosophy. Player-tracking cameras, installed league-wide in the 2010s, gave teams location data for every shot, pass and defender, letting analysts quantify not just where shots came from but how open they were and what actions created them. Shot-quality models now inform everything from play design to player development: young players are coached out of long twos almost from the moment they enter a professional programme. For readers new to this world, our beginner’s guide to sports analytics covers the foundational concepts, and the broader pattern across sports is mapped in the ways data and analytics are changing modern sport.
The backlash, from an unexpected source
Success bred sameness, and sameness bred complaint. Critics argue that when every team solves the same equation, the league converges on a homogeneous style: spread the floor, hunt threes and rim attempts, and let variance decide close games. The most striking voice in that chorus is Morey himself, now running the Philadelphia 76ers, who has publicly told the league office that the game has become unbalanced, saying of the three-pointer that “three is too much for that shot.” The league, for its part, has examined the data and concluded that no rule change is imminent, but the fact that the architect of the revolution is questioning its end state tells you how far the pendulum has swung. The aesthetic debate now sits alongside other analytics-driven controversies, such as the rest-and-availability fight covered in how player load management became a league-wide debate.
It is worth stressing what the analytics movement did not do: it did not eliminate skill, and it did not make the mid-range extinct so much as exclusive. Elite shot-makers still take and make difficult twos in the playoffs, when defences take away the efficient options; the difference is that the mid-range is now a counter, deployed by the few players good enough to beat the maths, rather than a default. Meanwhile, the same expected-value logic has spread to other sports, reshaping fourth-down decisions and play-calling in American football, as we explore in how data is reshaping football play-calling.
Where does shot selection go next? The frontier has moved from where shots are taken to how they are created: tracking models now evaluate screening angles, passing windows and defensive matchups in real time, and the next edge will likely come from optimising the possession before the shot rather than the shot itself. There is also a plausible future in which the league adjusts the geometry, whether through the arc’s distance or its value, to restore variety. But whatever the rulebook does, the deeper change is permanent. Basketball decisions, from the draft to the final possession, are now framed in expected value. The chart won. The question for the sport’s next decade is whether the game can stay beautiful while everyone is doing the same maths.
