IBM Knows What Makes Serena Williams So Good
It’s always hard to take your eyes off Serena Williams. But it’ll be especially tough at this year’s U.S. Open, where the tennis champ is currently working toward a single season Grand Slam. She’s just so darn good. But what is it, exactly that makes her so good?
Sure, we can all speculate—it’s her power, her serve, her stamina, the way she controls a point. But we can’t calculate precisely what makes her game so special. IBM believes it can.
Since 1990, IBM has been working with the United States Tennis Association to support the technological infrastructure of the U.S. Open. Back in the day, that meant generating scores and keeping the website up and running. Today, it means doing those things while also analyzing millions of data points about every player, every stat, every point, in every tournament, extending back for decades to derive insight about how a given match—or career—will play out.
The Next Serena
In addition to the U.S. Open, IBM now also works with the Australian Open, French Open, and Wimbledon. As this analytics operation has expanded over the years, IBM has created a rare window into not only which players are most likely to win, but why they’ll win, and what their opponents could do to change that. In other words, the data tells them what makes tennis players good. And that knowledge is becoming ever more important to the way we watch and understand the sport itself.
Take Williams, for instance. According to IBM, in an average tournament, Williams serves 65 aces—tennis lingo for serves her opponent doesn’t touch. As a result, she wins an average of 83 percent of the games she serves. Williams also runs drastically less than other female players, according to IBM, which captures player and ball position on cameras around the court. IBM calculates that Williams runs an average of 25.5 feet per point, compared to players like Garbiñe Muguruza, who run an average of 36.6 feet per point. And while her serve game is strong, her return game is too. In an average tournament, Williams wins 33 games served by her opponent.
But arguably more powerful than understanding Williams’ game is being able to apply that knowledge to all of the other female players in tennis to determine who might stand the best chance of becoming the next Serena Williams. That’s where IBM’s trove of data comes in handy. This year, the company filtered through the entire lineup of female competitors to find which ones, like Serena, have both a strong serving percentage and a strong return percentage, and landed on two players: CoCo Vandeweghe and Madison Keys, neither of whom are ranked in the top 10.
“Nobody has Serena’s return, but these two are the closest,” says Elizabeth O’Brien, who works on IBM’s sponsorship marketing team. “It’s about finding the levers where you can increase your percentage by 2 percentage points, 4 percentage points.”
This process can also unearth players’ weaknesses. For instance, a player’s second serve is often much slower than the first, because players are being cautious. IBM can look into how well that strategy plays out for any given player by analyzing how many points that player wins on his or her second serve. The company can drill down even further to look at how many of these points the player wins against opponents who have particularly strong returns. If the player is winning those points anyway, there’s no reason to shift strategies. If the player is not winning those points, there may be.
IBM can get even more granular, analyzing a player’s likelihood of choking when they’re down several points, or how their serve percentage changes when their opponent is one point away from winning a game. Already IBM has turned some of its basic analyses into tools for fans. Its SlamTracker app, for instance, breaks down match stats in real-time. It also rolled out a feature called Keys to the Match, which analyzes historical data to figure out exactly what it would take for one player to beat another player, taking into account both players’ strengths and weaknesses and past performance data.
These tools and others are being used by commentators, journalists, and to some extent, even the players and their coaches, who receive a USB stick of each match, complete with IBM’s analysis. But most of what IBM learns about these players happens in an ad hoc way, requiring a human being to come up with a question then search through the database for the answer. “Having that domain knowledge helps us figure out where to look for anomalies, and when we find anomalies, like an unusually slow average second serve, then we know where to run the query,” O’Brien says.
IBM’s hope, however, is to someday use its artificial intelligence tools like Watson to seek out those anomalies without human assistance. “It’ll be interesting as we continue to evaluate Watson,” she says, “If Watson can learn the questions to ask, and the systems are in place to answer those questions, it’s a virtuous circle.”
See original article here: