Skip to content ↓

Analyze this

MIT Sloan sports analytics conference reaches mainstream
Dallas Mavericks owner Mark Cuban, left, speaks with Houston Rockets General manager Daryl Morey MBA '00, right, during the MIT Sloan Sports Analytics Conference.
Dallas Mavericks owner Mark Cuban, left, speaks with Houston Rockets General manager Daryl Morey MBA '00, right, during the MIT Sloan Sports Analytics Conference.
Photo: John Marcus/
ESPN columnist Bill Simmons, left, listens to Jonathan Kraft, right, president of The Kraft Group, which owns the New England Patriots, at the MIT Sloan Sports Analytics Conference.
ESPN columnist Bill Simmons, left, listens to Jonathan Kraft, right, president of The Kraft Group, which owns the New England Patriots, at the MIT Sloan Sports Analytics Conference.
Photo: John Marcus/

Once, if you wanted to become the general manager of a professional sports team, you had to have been a great athlete. For decades, sports teams were almost exclusively run by former players.

Times have changed. Today, an MBA can be a route into the NBA. Take Houston Rockets General Manager Daryl Morey, who has the height and bearing of a basketball star, but never played professionally. Instead, Morey graduated from the MIT Sloan School of Management in 2000, and parlayed his analytical skills into his current job.

“All else equal, it is preferable to have played the sport,” Morey said on Saturday, during a panel at the fourth annual MIT Sloan Sports Analytics Conference, which he co-founded. But all else is not equal: Sports are awash in misguided conventional wisdom, and scores of former players have blatantly mismanaged franchises. So Morey is in the vanguard of general managers applying the analytical techniques of academia to basketball.

In practice, that means Morey’s staff has been dissecting the sport, doing things like pinpointing the most efficient shot location (the three-pointer from the corner), and slicing defensive performance into small, measurable elements, in an attempt to quantify how effective Houston’s players are. Their forward Chuck Hayes, for example, would be considered too small for his position, at a mere 6’6,” according to the conventions of coaches and scouts, but new-school metrics indicate that Hayes is a defensive ace. “You have to have a culture where there are no bad ideas,” said Morey, meaning he encourages his staff to develop new ways of assessing talent. As a result, a year ago, unheralded Houston pushed the eventual champion Los Angeles Lakers to the seven-game limit in their playoff series.  

To be sure, the field of sports analytics has existed for years: The baseball writer Bill James’ pioneering annual book, “The Baseball Abstract,” began reaching a national audience in 1982. The subject gained new popularity through Michael Lewis’ best-seller, Moneyball (Norton, 2003), which chronicled how the Oakland Athletics were using James’ principles to find undervalued players.

The Sloan conference, which featured panels examining analytical techniques, and research papers on subjects like blocked shots in basketball (not all of them are equally valuable), reflects this wave of interest. Saturday’s event, held at the Boston Convention & Exhibition Center, drew more than 1,000 attendees, up from 400 last year; half the NBA’s teams had a representative present.

Is Plus/Minus a plus?

The current state of analytics varies widely among sports. Baseball is the most developed, because it largely consists of a series of individual confrontations between pitchers and hitters, whose results can be easily isolated. As a morning session on “Baseball Analytics” made clear, defense is the last statistical frontier of the game, and even there, statistician John Dewan estimated, observers know “60 percent” of everything they can.

Baseball analytics are so thorough, “Now I don’t think you even have to watch baseball” to dissect it, quipped columnist Bill Simmons during an afternoon panel. Indeed, he added, you may not even “need to know how to hold a bat.”

But other sports feature the simultaneous interaction of many athletes at once. Isolating an individual’s performance in these sports remains problematic.

“Unlike baseball where you have a lot of discrete events, in football there is a lot of interplay, so it’s more difficult to analyze,” said Parag Marathe, a San Francisco 49ers executive, at a panel on “Emerging Analytics.” Consider a 25-yard run. How much of the credit goes to the running back, his blockers, or to defense lapses? “The NFL is a little bit behind” in analytics, Marathe suggested.

To work around the problem of complex interactions in basketball, analysts are refining the concept of “Plus/Minus,” which records how many points a team scores and allows when a particular player is on the court, per 100 possessions. One winner of the conference’s research-paper contest this year attempted to improve the concept; Dallas Mavericks owner Mark Cuban has tried to use Plus/Minus, while recognizing its flaws.

“There are all these qualifications you need to keep in mind,” Cuban said in an interview with MIT News on Saturday after he spoke on two panels. A player’s Plus/Minus can depend on the quality of his teammates, the quality of opponents, the tempo of play, and more. Currently, Miami’s Dwyane Wade leads the league in Plus/Minus, relative to how his team fares when he is not on the court, but that may just mean that he has worse teammates than Cleveland’s LeBron James.

That said, Cuban thinks the metric works well in evaluating the success of different five-man lineups, not just single players. “We’ve adjusted lineups in the playoffs based on our Plus/Minus numbers,” Cuban said. In 2005, Dallas lost the first two games of its first-round series to Houston, which was using a smaller, quicker lineup. The Mavericks studied the Plus/Minus numbers, reduced lumbering center Erick Dampier’s minutes, and rallied to win the series in 7 games.

“Mark Cuban helped break me out of that mold of looking at traditional statistics,” recounted Avery Johnson, the Mavericks’ coach at the time, while speaking on a “Coaching Analytics” panel. “Using Plus/Minus helped me out a lot in terms of my substitutions.”

Well, until the team faltered. In 2007, Dallas entered the playoffs with a league-best 67-15 record. But as Johnson recounted, the numbers showed that the Mavericks fared worse against their first-round opponent, the small-but-quick Golden State Warriors, with Dampier on the court. Johnson benched Dampier, the team’s starting center, for the series’ first game. “It was the right thing to do,” Johnson said. But his players did not like the adjustment; the Warriors quickly knocked out the Mavericks in a stunning upset.

The limits of metrics

As the Mavericks’ experience suggests, analytics have limitations. General tendencies may not be borne out in specific situations. Moreover, “I think there is an onus on whoever is dispensing that information” to explain it clearly and persuasively to everyone else, asserted Simmons, whose own recent tome, The Book of Basketball (ESPN 2009), mixes empirical data and subjective impressions while judging players and teams in NBA history.

And Morey noted another problem: In a business with short careers, changing circumstances may make some sports analysis irrelevant. “I think there are fundamental things that can be solved,” said Morey. “But by the time you have enough confidence in them, the world has changed.”

What has also changed, though, is that savvy sports fans now envision a future in the business. Take Matthew Martell, a senior associate at Octothorpe Software, a Vancouver firm that designs decision-making programs. Martell, capable of talking knowledgeably about sports-analytics problems in basketball, football, and soccer, made a 12-hour trip from British Columbia on Thursday, changing planes twice, to attend the event. “This is where you want to be, to meet and see the people who really know analytics,” said Martell. “It’s incredible to be here.”


Related Links

Related Topics

More MIT News