September 6, 2012

The Strange Case of Tennis Analytics

There are a couple sports where you don’t necessarily think of analytics. Tennis is one of them. This isn’t to say that statistics and numbers don’t play a critical role in the game, it’s just a different beast than baseball, basketball and football.

One difference is the variety of statistics. Unlike other sports, the statistics you see for tennis today are not all that different from the ones 30 years ago. The major stats you see on ESPN during the U.S. Open are still winners, first and second serve percentages and unforced errors. But the potential for understanding and using the data in different ways is still there.

As a result, many people say the ceiling is still fairly high for analytics in tennis. The MIT Sloan Sports Analytics Conference this past March had its first annual Tennis panel featuring former players Todd Martin and Paul Annacone as  well as US-based Australian coach Craig O’Shannessy and ESPN’s Marc Stein on the panel. Stuart Morgan, a performance analyst at the Australian Institute of Sport recapped the highlights of the panel on his blog.

“Tennis is like boxing. You don’t get an invitation to hit someone on the jaw, you have to keep jabbing away to keep your opponent from hurting you,” Martin said at the panel. “You wait for an opportunity to force your opponent into a position of weakness where you can strike them.”

Martin also said any raw data in tennis is useless without context. For instance, he says that players are not trying to win every point so analytics needs to be mindful of that subjective nature.

But how does technology and science try to incorporate a player’s tendencies into analytics. Well they have to understand them first and that’s where IBM comes into play. This year, IBM has been trying to incorporate their analytics software to better understand the best tennis players in the world. They started with something called Slamtracker according to InformationWeek.com:

SlamTracker is an online dashboard that serves up match-by-match analysis based on seven years’ worth of Grand Slam data, which translates to about 39 million data points. The system mines data on match outcomes by player, aces, faults, errors, playing surface, and more, then models the patterns likely to emerge in a pending matchup.

“It’s not a prediction who will win; it’s a look at what each player needs to do well to have a higher likelihood of winning,” explained John Kent, program manager of IBM Worldwide Sponsorship Marketing, in an interview with InformationWeek.

The system’s Keys to the Match analysis might reveal that a player will need to win more than 72% of points on first serves, convert more than 60% of break point opportunities, or win more than 40% of first-service return points. Djokovic, for example, managed to win more than 48% of three-to-eight-shot rallies against Nadal at the French Open, as SlamTracker prescribed, but he failed to win more than 44% of first-serve return points.

image

But to really get into the specifics of a player’s movements and what they can do, IBM is has introduced something called SecondSight to tennis. It’s similar to the technology the NBA uses to show shot selection, rebounding, and positioning. The MIT Sloan Management Review discusses the new technology:

Every serve, smash, swish and slam is being monitored, measured, analyzed and reported — in real time. The All England Lawn Tennis and Croquet Club (AELTC), in conjunction with IBM, has built two big data analysis systems for tennis that are designed to provide detailed player and tournament insights, and historical comparisons.

IBM SecondSight is still in trial phase. It was used last year on Court 18 to track how players move on the court. This year, SecondSight is moving on up to Centre Court. Using 3D cameras initially developed by the military, it measures and tracks balls and players in any dimension. SecondSight adds real time analytics from that 3D play to develop game insights. The goal, says IBM: add a new dimension to understanding the science of tennis.

So while context is obviously important when it comes to understanding all sports and statistics, it is absolutely critical when it comes to tennis. It is a typical example of how numbers alone cannot tell the whole story.