While there’s no real gauge for how strong an idea can be, a decision by Booz Allen Hamilton to build a practice around that idea might be a good indicator of longevity. That is good news for the sports world, or at least the companies that want to cash in on it by catering to owners’ desires to win and leagues’ desires to attract more fans. Because the mega-consultant to the U.S. government and the world’s largest companies now thinks it can make money applying its knowledge of data science to the world of professional sports.
We’re not talking about Moneyball, either. Rather than analyzing player statistics to try fielding the best team at the lowest cost — something that’s all but common practice at this point — Booz Allen wants to help teams (and potentially their fans) understand what’s happening on the field. That means analyzing everything from who’s where on a basketball court to how hard a football player just got hit, to help figure out how it affects outcomes and injuries.
Granted, it’s more of an experiment right now than a full-fledged business line, but Booz Allen does see a future in sports. “We don’t have any [public sports] big data engagements that we’re actively working against right now,” said Ray Hensberger, a senior associate in the company’s Strategic Innovation Group, where the idea blossomed as an internal R&D project around trying to predict the next pitch in a baseball game. But the company is making connections and he’s confident it will attract a lot of clients.
Sensors, sensors everywhere
If you attend a conference like the MIT Sloan Sports Analytics Conference, like Hensberger did in early March, it’s easy to see why someone who’s versed in analytics would be excited. There are sensors for measuring everything and they can go anywhere from shoes to mouthpieces. Someone — teams, television stations or the leagues themselves — is recording every movement. Once teams get used to collecting the data, Hensberger predicted, the next thing they’ll want to do is analyze it.
This goes even beyond stuff like Booz Allen’s work on predicting pitches or other efforts to quantify whether punting in football games is a good idea. Hensberger thinks it’s possible to predict injuries by analyzing how athletes are contorting their joints during games, or to determine the effect of sweat levels on how well a curveball curves. Being in the zone is a real thing, he noted, and even though football is the ultimate team game, the addition of a single player like Peyton Manning can elevate an offense into the stratosphere.
“The question is why, and how do you actually measure that?” Hensberger said.
Already, certain leagues are trying. The National Basketball Association, for example, is using a system called SportVU from a company called STATS LLC to tracker the location of every player on the court at every moment during the game. STATS LLC analyzes this to produce a number of statistics itself (relatively simple things like “touches” and “catch and shoots”), but that’s the tip of what’s possible with the massive datasets its software create.
However, moving from statistics and simple visualizations to predictive models is a big step — and that’s where the data experts come in. This blog post from Fathom Information Design shows what’s possible in tracking player movements using the SportVU data. This story in Grantland discusses a new measure called “expected possession value” that its creators claim can predict the resulting points at any given time in a game depending on who has the ball and who’s where on the court.
Professional racing teams are already well known for how they measure every little aspect of their vehicles and races to achieve even the most-minute improvements in speed and handling. At our Structure Data conference, now less than two weeks away, McLaren Applied Technologies’ Geoff McGrath will talk about the Formula One team’s impressive strategy of combining math, data and simulations to influence the design of racecars, racing bikes and even data centers. (We’ll also RunKeeper CEO Jason Jacobs talking about personal exercise tracking, and Peter Guerra from Booz Allen Hamilton talking cybersecurity.)
Resistance is futile
But Hensberger thinks even more-traditional leagues such as Major League Baseball (which is stats-heavy, but slow to adopt things like instant replay) and the National Football League (which still relies heavily on coaches’ knowledge rather than data for in-game decisions) will come around if (A) it works and (B) it stands to make them money. That doesn’t mean coaches will be irrelevant, it might mean teams will opt to “let computers crunch over all this data and then let humans make the final decisions,” Hensberger said.
Although, Hensberger noted, money doesn’t have to come just from winning. It might, for example, come from new ways of delivering real-time experiences to stats-crazed fans, whether that’s via traditional media like TV or new media like Google Glass. Advanced data analysis might also affect the flow of money between players and clubs when it comes time for salary negotiations.
Hensberger is a realist, though, and understands that large-scale implementation won’t happen overnight. “‘Are we playing it the right way?’ ‘Are there better ways to optimize the teams and the league overall?’ That’s a scary thing to some people,” he acknowledged. But ultimately, he added, “I don’t think there’s any way to ignore it.”
Feature image courtesy of Shutterstock user Anthony Correia.