Every step they take

IN THE late 1990s Billy Beane, the general manager of the Oakland Athletics, a baseball club, made a radical change to the club’s recruiting methods: he decided to apply some science to them. Instead of relying on the instincts of scouts, he and his deputy Paul DePodesta, a statistically astute Harvard economics graduate, crunched candidate players’ numbers. Baseball is awash with such numbers, and the two men applied simple statistics to them to identify valuable players whom scouts had rejected, and who could thus be hired cheaply. The result transformed the club’s performance. Despite having one of the league’s lowest payrolls, it qualified for the post-season tournament run by Major League Baseball (MLB), the professional game’s organiser, for four years running.

This success, which was the subject of a book (and later a film) called “Moneyball”, caused others to copy the method. Before long, teams had extracted all the information they could from the game’s traditional statistics. To produce even more accurate predictions, they would need better data. And the means to gather those numbers have now arrived, in the form of a system called Statcast developed specifically for baseball by MLB.

Statcast can, pretty much, follow and record everything that happens in a baseball game. It builds on earlier game-tracking technology, such as the Hawk-Eye system used in cricket, but is far more sophisticated. It constantly logs the position of the ball and of every player on the field. It calculates the speed and curvature of a pitch, how rapidly the ball spins and around what axis, and how much faster or slower than reality that pitch appears to be to the hitter, based on the length of the pitcher’s stride. When the ball is hit, the system measures how quickly it leaves the bat and how its path is affected by atmospheric conditions. It then tracks how long fielders take to react before moving, and the efficiency of their routes to the ball’s eventual landing spot. And it takes just 15 seconds to crunch these numbers and integrate them with video recordings.

Statcast captures the information it needs by fusing data from two pieces of equipment. One follows the players. The other follows the ball. The player-following system is a stereoscopic camera array developed by ChyronHego, an American graphics company. This sits behind third base and takes 30 snapshots a second. It cannot, however, track the ball as reliably as it tracks the players, for baseballs (which are white) are hard to see when they fly in front of similarly coloured backgrounds. To follow the ball, including measuring its spin using the Doppler effect (which causes the radar beam’s frequency to rise when it bounces off part of the ball that is spinning towards the detector, and fall when it returns from part that is spinning away), the MLB therefore turned to TrackMan, a Danish radar firm. Their system had a hiccup when it turned out that the giant video screens without which no major-league baseball ground is complete were jamming it. But that has now been resolved.

In theory, following the players and following the ball in this way should provide all the data needed for statistical purposes. But Statcast is designed, in part, for television, and for that purpose yet another camera needs to be added to the mix. This watches the field as a whole, providing co-ordinates that map the radar and optical data onto broadcasters’ video feeds. And the result does indeed make for compelling TV. It permits commentators to illustrate replays with dazzling visual displays. It was used this way for the first time on April 21st, by MLB Network, the sport’s proprietary cable channel.

It is, though, the trove of information Statcast’s all-seeing eye delivers about how players play that most interests teams. They can use such data to make players better—for example, by telling a pitcher who cannot impart sufficient backspin to try a different grip—and to allocate resources more efficiently, such as moving a fielder with lightning-quick reflexes to third base, a position where such people are particularly valuable.

Teams can also use Statcast to help with recruitment. In 2013, for example, the Houston Astros employed an analysis based on the TrackMan system to acquire an unaccomplished pitcher called Collin McHugh, because of his fast-spinning curveball. They then told him to throw that pitch far more often during the next season, and he blossomed into a star.

Who will have the opportunity to dig into Statcast’s numbers does, though, remain to be decided. When a prototype version was launched in 2007, MLB allowed anyone to download the raw data. That open-source approach led to a flurry of discoveries, many of which have become accepted wisdom. It also helped teams identify and hire promising analysts. Now, those teams with good analysts have an incentive to pull up the ladder. They fear weaker rivals will free-ride on the work of the next generation of amateurs, reducing the edge that employing professionals gives. Rob Manfred, MLB’s commissioner, has promised fans “very good access” to Statcast. What that actually means remains to be seen.