The stats revolution isn’t just for baseball anymore
Author Joe DeLessio
A little more than a decade ago, Michael Lewis turned the baseball world on its ear with the publication of Moneyball. Lewis’ best-selling book told the story of how the data-driven Oakland Athletics’ front office continually put forth a winning team despite having much less money to work with than its competitors, and it introduced the work of pioneering analyst Bill James to a wide audience. Back in 2003, when Moneyball hit shelves, a stat as simple as on-base percentage wasn’t appreciated by many Major League Baseball teams. In the years since, statistical analysis has become so mainstream that you can find an advanced stat like WAR (wins above replacement, or the number of wins a player is worth to his team) on the back of trading cards; writers like Nate Silver, who got his start at Baseball Prospectus before he became an election-forecasting oracle, have gone mainstream; and in 2011 Brad Pitt starred as Oakland General Manager Billy Beane in the Oscar-nominated film adaptation of Moneyball.
Roughly a decade after Lewis published his groundbreaking book, the stats revolution that changed baseball is now beginning to take over other sports. This development may surprise some, but Baseball Prospectus editor in chief Sam Miller says it’s something baseball analysts “absolutely” expected. “Once you read Moneyball, you saw the possibility in everything,” he says.
Take the NBA: Once upon a time, the only stats captured in a pro basketball game were the ones that appear in the box score—things like points, rebounds and assists. Then, former ESPN writer and current Memphis Grizzlies executive John Hollinger developed a stat called PER, or player efficiency rating, which uses a formula that incorporates many existing stats—including shots made and missed, rebounds, assists, steals and turnovers, among others—into a single number to quantify a player’s worth. The stat reinforced some things we already know (last season’s two top-rated players were Kevin Durant and LeBron James, with PERs of 29.9 and 29.4, respectively) and told us some things we may not have (second-year New Orleans forward Anthony Davis should already be considered a superstar, as his 26.54 PER ranked fourth in the league).
And the NBA is taking things even further. In every arena last season, the league installed a camera system called SportVU, which tracks the movement of the ball and every player on the court to produce a much richer set of data. For example, SportVU tracking tells us that in the 2014 playoffs, San Antonio guard Tony Parker traveled the most distance and touched the ball more than any other player, and that Oklahoma City guard Russell Westbrook scored the most points on both drives to the rim and pull-up jumpers. Teams can then break down such data to better assess not just their own players but opponents, too.
“It was always kind of an inevitability,” says Alex Rucker, a senior analytics consultant for the Toronto Raptors, of the league’s embrace of advanced stats. “Data-driven analysis is clearly a good thing, pretty much across the board. Making decisions that are informed by objective data yields better results than making decisions based solely on hunches or guts or experience.”
Rucker, who notes that every team in the NBA has some analytics presence in its front office, says that all this new data can provide insight into things like player defense—something that couldn’t be done with basic, old-school stats, or even PER. For example, SportVU shows that Indiana center Roy Hibbert holds opponents to a lower shooting percentage near the rim than his shot-blocking contemporaries do.
NFL teams, too, are embracing statistical analysis. “I think that they’ve been more and more enthusiastic over the last few years,” says Aaron Schatz, editor in chief of Football Outsiders. “There was a definite turning point last year, at the [NFL Scouting] Combine, where suddenly analytics was a watchword for a number of teams to talk about things they were trying to do.”
Unlike in the NBA, where the 30 teams banded together to install the SportVU cameras and work with the same data (albeit in their own ways), everything NFL teams do is proprietary and heavily under wraps, but franchises like the New England Patriots and the Philadelphia Eagles are known to give particular credence to data-driven analysis.
The fact that these stats are advanced doesn’t mean that they have to confuse matters. In fact, Schatz says that much of the advanced statistical analysis in football has to do with identifying “nonsense” and providing context for what happens on the field.
For example: It’s common to talk about a running back’s total yardage, but it’s preferable to consider how successful his runs were, factoring in the down and distance of the play. A five-yard run on third-and-four, after all, is much better than a five-yard run on third-and-six. One of the stats developed by Football Outsiders, DVOA (defense-adjusted value over average), measures a team’s success by looking at every play in the league and assigning it a “success” value based on yards needed and yards gained. The site then compares a team’s performance to the league baseline. It shouldn’t come as much surprise that the Seattle Seahawks, who won Super Bowl XLVIII in a blowout, had by far the best DVOA in the NFL last season.
And that’s the thing: While a stats-versus-old-school debate consumed baseball for years, smart analysts acknowledge that an either-or approach is limited, and that scouts and number-crunchers should coexist. This is very much the case in the NFL, according to Schatz. “You didn’t develop this dichotomy that you had in baseball, where it was like the stat people versus the scouting people,” he says.
Of course, the rate of acceptance varies from sport to sport. Statistical analysis is still gaining a toehold in NHL front offices, but the Chicago Blackhawks, who have won the Stanley Cup twice since 2010, embrace the use of advanced stats in their evaluation of players. Teams are increasingly staffing their front offices with data-savvy executives who aren’t afraid of metrics like Corsi (which uses the number of shot attempts by a team or player as a proxy for puck possession) or PDO (which looks at shooting and save percentages against a regression model to tell you if a team or player has been particularly lucky or unlucky). Just this summer, the Edmonton Oilers hired Tyler Dellow, a former hockey blogger whose site was a trove of data, to work in their front office.
The stats revolution is only gaining steam. Rucker says that he’d be stunned if, 10 years from now, every NBA team didn’t have four or five statistical analysts. “I think those positions will become increasingly technical—more computer science and more engineering and less statistics, just because of the kind of big data world we’re diving into,” Rucker says. “Among the non-baseball sports, it’s this really kind of cool, interesting time, where there’s an opportunity for tremendous growth.”
Joe DeLessio covers sports for New York magazine’s website. He’s glad he took that statistics class in high school.