A couple of Tennis systems

[Joe]

These both make use of the favourite-longshot bias (now well established I think), and more significantly in the first, where pricing is the only criteria for bet selection. I've now got odds data back to Jan 2001 (2 and a half years or about 7,500 matches). Up to 6 bookies, although never more than 5 at one time (because of what I've collected). System 1 Back any maximum price when both: a) maximum/average price ratio is greater than 1.02 b) average price is less than 1.25 Results: 747 bets 24.28 unit stakes profit 3.25% yield Time series: The reasoning behind this system is that the average price, when below 1.25 will be pretty close to the value line because of the bias, so backing prices greater than this should show profit. In fact average max/average price ratio was 1.035. If we increase the ratio for criteria a) incrementally from 1.02 through 1.03, 1.04, 1.05 etc., the average max/average ratio rises nicely in line with the yield (although of course the number of selected bets falls), thereby providing further confimation that the average price (when below 1.25) is about on the value line. Make sense? For the record, average (maximum) price of these bets was 1.22 System 2 Back any maximum price when both: a) the difference in ATP entry ranking between 2 players is 100 or more b) the ATP entry ranking of the favourite (in terms of ranking, not odds) is 100 or lower. {For these series of bets I took out about 5 matches involving Krajicek after his return from injury when his ranking was 1093. He's a quality player and the system obviously broke down here, as evidence by his results for this tournament.} Results: 997 bets 31.68 unit stakes profit 3.18% yield Time Series: I was quite surprised to see this thing work out since the ranking does not take into account different surfaces. Obviously some of these bets were also in the first system, but only 241 (for which a yield of 7.45% was returned when applying both sets of criteria to bet selection). Average (maximum) price of these bets was 1.36 Lumping both systems together (and placing 2 bets, i.e a stake of 2, where both systems pick out a match) returned 55.97 points from 1744 staked, or a yield of 3.21%. With 5% current bank staking, a 100 starting bank would now be 900.

[Joe]

Re: A couple of Tennis systems For the record, backing every average price 1.8 or shorter returned a -4.4% yield. Backing the longer prices returned -21.3%