Home-court in tennis: Update
In yesterday’s post, I presented some research on the degree of home-court advantage in men’s tennis. Much of my reasoning was correct, but my dataset was flawed.
My intent was to exclude matches in which at least one of the players received a wild card into the tournament. This is an obvious selection bias when it comes to home-court: Wild cards are generally local heroes; they are also often players who tournament organizers think are better than their rankings reflect.
In processing the data, I got rid of some of the WCs, but nowhere near all. Of my sample of 666 matches, 222 (yikes, this is getting creepy) had a wild card in them. So that leaves a sample of 444 matches from 2009.
Of the 444 matches, the home player won 266, or 60 percent. Using my algorithm for predicting the outcome of each match, the home player should have won 227–just barely more than half.
In other words, my surprise yesterday (and ensuing convoluted explanation) was purely a result of including the wild cards. On average, the home player and his opponent were equally matched.
The new conclusion is that players are about 17 percent more likely to win when they are in front of their home crowd and their opponent is not. To include this in my algorithm, multiply the home player’s ranking points by 1.5.
A note on wild cards
We can draw some obvious conclusions about wild cards, too. If removing WCs from the sample means the home-court advantage goes down, the home-court advantage for WCs must be bigger. And since the majority of WCs are playing in their home country, we may be able to say that there is some sort of “WC advantage.”
As I’ve suggested, it isn’t so much that wild cards have an advantage, it’s that a wild card’s ranking points are much less likely to accurately reflect the player’s skill level. To take just two examples, David Nalbandian and James Blake both received several wild cards this summer. Both players are returning from injury. My algorithm sees them as players outside of the top 100, but obviously, when healthy, they are far better than that.
(That’s one reason I’m interested in the notion of ‘peak ranking.’ If Pete Sampras came out of retirement tomorrow, he’d almost instantly be able to beat players in the top 50, if not the top 20 or even top 10. But it would be some time before his ranking reflected that. We’ve seen this phenomenon in the women’s game with Kim Clijsters and Justin Henin.)
There’s a lot more to study here on another day.