Most bettors are betting on the tennis matches at the Australian Open 2009 nowadays. I came across a publication which discusses about forecasting the winner of a tennis match and I back tested the suggested mathematical formulas using the historical data from the previous Australian Open Championships. While this is the first formula I found to make use of the world tennis ranking system, the results in my Excel spreadsheets are disappointing.

The publication about forecasting the winner of a tennis match is 11 pages long but you don’t need to read them all. The mathematical formula we are interested in is in the 5th page and is the (4) formula. I went on and created the Excel spreadsheet where you can input the ranking of two tennis player competing in the tennis match you are going to bet, and the sheet will automatically calculate the estimated probability the highest ranked player to win the game.

You can download the Excel file here.

For back testing purposes I used the invaluable data found at Tennis-Data.co.uk and I considered the Australian Open statistics since 2006 season. So, betting in 400 games in the Men’s Tennis Championship during the last 3 years using the formula above, would lose us a lot of money.

back-for-ranking

If you think that by backing the opposite outcome will make you money, guess again. Commission is the big winner here, since we still lose money by backing the other tennis player, betting as if the formula picks out the loser. I should note here that I took into account the average bookmakers’ odds and I applied 5% more commission on them, making it a bit more expensive.

 back-against-ranking

However, if we lay the initial selection, we will make money. How’s that possible? Well, when I backed either tennis player, I used a fixed amount of money to risk on each bet, €10. When laying I aimed for a fixed amount of money to be won instead of risking, again €10. In this situation, there are times that I risk a lot more than €10 since laying at odds over 10, we risk more than €100 to win €10. Considering that the sample was a bit small and just a few big outsiders won a couple of games, the graph is justified. 3 or 4 losing bets of 15.0 or bigger odds would wipe out all my winnings.

 lay-against-ranking

Nevertheless, if you were searching for a formula to calculate probabilities from the tennis world ranking system, you may give it a go and see for yourself. A final warning if you are going to test it in the women’s championship, as the “λ” constant needs to be changed to 0.7150.

And if you happen to make some money using a strategy based on this equation, don’t forget to leave a comment.

Get my Posts in Your Email For Free!

I will deliver my posts straight to your inbox as soon as they're published.

  • Jason

    Hi Jim

    I found your website last night, the content here is excellent.

    I have two questions:

    1.Of your sample of 400 matches, was each bet/lay placed on the basis of value? ie were the average odds offered incorrect to offer a ‘profitable’ scenario, and if they weren’t did you not bet on the match?

    2. Is laying at 10.0 odds (laying €100 for a €10 win) mathematically the same as backing at 1.10 odds (betting €100 for a €10)?

    Kind regards,

    Jason

    • Hi Jason, thanks for your comment and nice words.

      1) As far as I can remember, yes, only value bets were considered. Since I always compare the calculated probability percentage with the actual odds, I’d be surprised if I didn’t do that here as well. So, in brief, yes, I would only bet at value odds and would ignore non-value bets.

      2) Quite close, since by laying 10 euros at 10.0 our liability is in fact 90 euros, not 100. But yes, it would be about the same as betting (backing) 100 euros at 1.10 odds, since we risk 90-100 euros to win 10. Actually backing at 1.10 for 100 euros is the same as laying 10 euros at 11.0 (liability: 100 euros).