Wednesday, May 31, 2017

Statistics and Sport

We all know that statistics don't tell the whole story, but sometimes they can fall into the "quite interesting" category and even make make to the status of "very interesting"! Take the Aviva Rugby Premiership Final for example. Exeter ran further and passed more. They had more possession and more territory than Wasps (68% to 31%). And yet the game was decided by a single penalty.

Dig a little deeper and you find that Wasps made more line breaks (15-8), but everything else was pretty similar in attack. Defending they made more tackles (282-120).

So we might say that Exeter's victory is reflected quite clearly in the statistics. But that's not the whole story. Wasps were pretty shambolic in the fist half and in the end their lack of precision and accuracy probably cost them the game. That and a certain forward's reluctance to let go of the ball when the referee told him four times to get his mitts off it! And what if Exeter hadn't scored a late penalty to level the scores? Wasps would have won the game having had half the possession and half the territory Exeter had. Interesting.

So are the statistics ever useful, particularly when it comes to coaching? Clearly from a rugby point of view the obvious thing the statistics tell you is that it's pretty difficult to win a match when you spend most of it defending your own try line as Wasps had to do on Saturday. But then again that was pretty obvious from simply watching the game.

But what about other sports, particularly the one in which I am most involved, tennis? There's a much vaunted statistic often discussed in post-analysis by commentators and pundits. It's the winner:unforced error ratio.

I watched some of the Murray/ Kuznetsov match yesterday and I had a look at the stats from it. Murray's ratio was 29:24 (1.20), Kuznetsov's was 33:45 (0.73) But does that tell the whole story? The thing about tennis is that you have to win more games than your opponent and that, you'd expect, means winning more points. But how many more? Take a really simple scenario where the match score is 6-4,4-6,6-4. You could actually win that match by winning only 2 more points than your opponent, the difference between wining and losing a single game. It could even be a single point and possibly no different at all although I'd have to do a lot of probability maths to work of if that was possible!

In the match Murray won a total of 118 points over 34 games. He won 22 games in total. He hit 29 winners. That's less than 30% of his points were won with outright winning shots. Now this is clay, and hitting winners on clay isn't easy! According to the stats, Kuznetsov made 45 unforced errors (not a helpful description because unforced is a rather subjective measure in my opinion). That leaves 44 points unaccounted for unless they are deemed forced errors, but that didn't appear in the statistics I was looking at. If you take both players together then out of 200 points played 63 were decided by winners (interestingly Kuznetsov hit more winners than Murray), whereas 69 were decided by unforced errors and the rest, 68, were unaccounted for in terms of how they were determined. So in this match everything seemed quite even and Kuznetsov's higher error count appears to have been a deciding factor.

What statistics can't tell you is how the style of play and the court surface, weather conditions etc affect the course of a match. An aggressive player may make more errors than a defensive player but their attacking style might create more opportunities to make winning shots. Two attacking players might have shorter rallies, two defensive players might have longer rallies. How might that affect errors and winners? Statically the player with a ratio higher than 1.0 should win, and that is nearly always the case. Nearly, but not always. On the ATP tour around 40% of matches are won by the player with the better than 1.0 ratio. On the WTA is nearer 90%.

So how does this affect the way we coach? Do we simply tell our players to make fewer errors, be difficult to beat, or do we need to adapt our coaching to their style of play. An aggressive all-court player might make more errors, but they might hit more winners too. Perhaps players need to learn how best to create the opportunity to hit the shots they hit best without worrying too much about error counts and winner/error ratios.

The overall stats from the ATP/WTA tours suggest that even if you keep the errors down and the winners up, you don't always significantly improve your chances of winning. Rugby and football matches can be won form statically weak positions. In then end, it just goes to show that sport is more than a statistical numbers game!

No comments: