# A mathematical equation to predict results of tennis matches.

#### Polaris

##### Hall of Fame
This is weird, but some of you might find it interesting. Swinburne University mathematicians have come up with a formula for predicting the winners of matches and tournaments. The article says that it has been used by some bookmakers. Here's the link:

http://www.timesonline.co.uk/article/0,,2087-1669340_1,00.html

Since it relies on statistics, I guess that, with enough data, it may be relied upon to predict the winner on average. But it will be interesting to know how much noise can throw off the prediction, i.e., how much worse should player A serve and/or how much better should player B return to reverse the prediction ?

Also, the formula is an unlikely and cool candidate for a signature on a tennis message board .

##### G.O.A.T.
just a quick comment....there are some flaws in my opinion in this theory and the article contradicts itself. first they say, the game is dominated by the serve (which it isnt), and then they say this <snip> At the Australian Open, which is played on hard courts, 61.7% of points are won on service. But on the grass of Wimbledon this rises to 63.8%. <end snip> if the game is so dominated by the serve, wouldnt the % spread be much bigger than 2.1% considering the Oz is played on the slower rebound ace and W is played (supposedly) on the best surface for the serve. i realize they slowed the grass down and the bounce is higher. didnt read past page1 but this seem contradictory IMO.

#### Kobble

##### Hall of Fame
This reminds me of one I was going to do. I realized that the statistics I needed were hard to come by, and equipment(racquets, balls, surfaces) is constantly changing. Plus, I don't know of a way to make any great money at it, so I dropped the idea. Maybe this will motivate me again.

#### West Coast Ace

##### G.O.A.T.
if the game is so dominated by the serve, wouldnt the % spread be much bigger than 2.1% considering the Oz is played on the slower rebound ace and W is played (supposedly) on the best surface for the serve. i realize they slowed the grass down and the bounce is higher. didnt read past page1 but this seem contradictory IMO.
I wonder if indoor carpet is now the best surface for big servers - by % of serve points won.

There's a chance that a formula could be developed - it wouldn't be 100% but if you could get to the >70% range you would make some money. Where does one get all the match data to validate against?

#### driger

##### Banned
we need an equation to tell us federer is the favorite? heres another formula: win - loses = x

##### G.O.A.T.
could be, but do they even play indoor carpet anymore? in anycase, an indoor hardcourt is also a servers dream, but the advantage of traditional grass isnt so much the speed it's how the serve skids and shoots thru the court on the grass. the new wimby grass seems to be alot higher bounding. it would be interestng if they did the same analysis with return of serve % won and see how well you can predict wins based on that

#### Morpheus

##### Professional
It looks to me like what they are saying makes perfect sense. Those with the best percentage of points won to points lost on their serves win more often. If you go to deuce/add frequently, you are probably broken more than if you always win 40-0 or 40-15. I can't see how anyone could argue that the game is not dominated by serving when players win most their games when its their turn to serve.

The statistics regarding grass suggest that serving is more important to winning than on other surfaces. That is hard to argue.

I believe their theory is most applicable to fast surfaces (or low/poor bouncing surfaces like grass) and is least predictive on clay.