Numbers Game
By TennisWorld Contributing Editor Andrew Burton
Morning, all.
Michael Lewis has a new book out on the recent - um - hiccup in the financial markets that Nearly Destroyed World Civilization As We Know It. Early reviews are very positive, which is no surprise: Lewis may be the best non-fiction writer currently working in the English language (feel free to make other suggestions, though). He has a gift for communicating complex ideas with vividness and clarity, and he makes it look absurdly easy while doing so.
Lewis writes primarily about mavericks - people on the outside who are not only just a little bit smarter than the rest of the herd, but have the guts, the moxie or the sheer bloody mindedness to act on their view when the rest of us, especially the comfortable elites - the Club - don't know what they're talking about. In Moneyball, the Club is made up of baseball scouts, managers and general managers who operate on the basis of received wisdom from the age of Babe Ruth. Moneyball's protagonist, Billy Beane, is the GM of the Oakland As, a team with a low budget in an era when the received wisdom is that big spending wins pennants. Beane turns to numbers - data, statistics, analysis, and testing - to unearth players who are worth much more to the club than their lowly salaries might indicate, and along the way dispenses with star players whose value to the team, calculated properly, is much less than their ginormous contracts:
What earthly relevance to tennis does this have, you ask (if you haven't jumped directly to the Comments to check on how Kleybanova is doing)?
Simply this; tennis is ripe for a revolution - the same kind of revolution that is sweeping through other fields of human activity like finance, medicine, electoral politics, and retail sales (think WalMart). Each of these activities is being transformed by data mining - the ability to find hidden patterns by accumulating large volumes of data and then subjecting them to statistical analysis and visual display.
Television viewers, and subscribers to services like TennisTV, have gotten a hint of this through visualizations created by Hawkeye. To most fans, Hawkeye performs just one task - it assesses where a ball landed and tells the chair umpire (via a visual representation of the ball's trajectory) that Roger Federer was mistaken again and the ball did indeed land out. But that's just a fraction of the information Hawkeye gathers during a match. From time to time, commentators will put up graphics showing where a player is hitting to (68% of balls going to Federer's backhand) or a player's hit point relative to the baseline - and they may compare how this evolves during a match, illustrating how a player is adjusting his or her game plan.
Even these graphics, whizzy as they are, understate what's possible. Just as a casino can use data mining to work out which blend of incentives will encourage more females over the age of 65 to play higher cost slot machines, an enterprising tennis coach, one day soon, is going to start trawling through the data to find out which shot patterns are most effective, against whom, and when. When you push Andy Murray wide to the ad side, how often does he slice cross court? How often does he go down the line? Does this change in key points, or as the match goes on? How heavy is the spin he puts on the ball, and which tactics, based on his last 100 hundred matches, have proved most effective against that shot choice?
If you're still not convinced, consider this: the majority of tennis points are short, and even those that last more than four or five shots usually resolve themselves into fairly straightforward elements - for example, CC BH rallies. Data mining is designed to find patterns hidden in large amounts of noise, but it works particularly well when the same elements are repeated thousands and thousands of times.
I'm expecting this suggestion to be fairly controversial, because for many of us tennis is about the combat of the human spirit, and who'd want to watch if that was eliminated and replaced by number crunching. But let me answer that line by suggesting that this is no different to leaps made in the history of the game - for example, with Ivan Lendl and Martina Navratilova taking athletic conditioning to a new level. The tennis stars of tomorrow will still need their legs and their lungs and their heart to play the game, but they'll also be getting some help from silicon.
So here's a fearless prediction (although it comes with a longer time fuse than Steve's ill-fated picks for IW); by the end of this decade, at least one player will have achieved results far ahead of his or her apparent talent or skill level, because their shot selection gives the impression that they know what their opponent's going to do before the opponent does. And they will, because someone with a laptop in their entourage will have literally done the math.
Earlier today I visited with one of the lady volunteers, Cheryl Smith, who records the shot and point statistics here at Indian Wells. Cheryl is an umpire and a 5.0 level player and coach, and she talks about what goes into doing her job in a YouTube video. So the next time you grumble about unforced error stats, you can be assured (if its at IW) that the scorers know what they're doing.
In other news, Marcos Baghdatis just knocked off the top seed, Roger Federer, 5-7 7-5 7-6(4). Doubtless you know that already.