As I wrote last week, tennis needs more detailed statistics. Most of all, we need them in an open format so that researchers can utilize all the data stored for every match. No use in have Hawkeye cameras on every court if the data stays locked up.
I’m working on a system for charting matches and storing extremely detailed serve and shot information. It will have to stay under wraps until I get a few more kinks worked out, but in the meantime, I want to show off some of what it can do.
Click here for more exhaustive serve data than you’ve probably ever seen before.
Today’s match wasn’t the most gripping that Roger Federer (or Grega Zemlja) ever played, but there’s still plenty of interesting stuff:
- Roger won 85% of first-serve points. No surprised there. More impressively, he won 60% of his first-serve points on or before his second shot. (That’s “<=3W” in the tables.)
- Fed went down the T with just under half his first serves (47%), but up-the-middle offerings accounted for 11 of his 12 aces.
- Zemlja hit a shocking 27 serves into the net–almost half of his faults, and just over 20% of all of the serves he hit today. (Watching the match, it felt like even more.)
- Roger’s first serves were somewhat more dominant in the deuce court, as he lost only three first-serve points in that half, and won two-thirds of his first-serve points in the deuce court by his second shot. In the small amount of data on offer today, he was noticeably weaker with his deuce court second serve, losing 5 of 12 second-serve points in that direction, compared to only 3 of 18 second-serve points to the ad court.
- Zemlja fared better serving to the ad court today (64% of service points won to 56% in the deuce court), and was particularly deadly when he landed a serve wide in the ad court. He won seven of the eight points that started that way, five of them with or before his second shot.
(If you didn’t click on the link the first time you saw it, now would be a good time.)
You get the idea, I hope. With this much data, the sifting is as important as the collecting. There are hundreds of data points we can generate just from tracking each player’s serve performance, and we can expect that most of them won’t have much to tell us.
And, of course, one match is just that–a small sample, fewer than 100 service points for each player. While we can look at these tables and gain some insight into exactly how Roger was dominant today, it would be a mistake to draw much in the way of broader conclusions.
For that, we’ll need more matches, more data. We’ll get there.