I have been backing and laying my AE ratings for flat handicaps since January 2014 and so far so good. Off a huge amount of betting activity my return on capital employed has been 166% with 3 months of the year remaining. It has been a great learning experience and as usual there have been drops and lengthy periods of stagnation. However I am not blogging simply to boast but to mention an interesting conversation I had with Mr FlatStats.co.uk. I am sure if you are a ratings producer you are always looking to improve your ratings even if its only a minor tweek. My conversation with Phil at Flatstats concerned whether one should alter all the SP’s on races to remove the overround when calculating AE values. Perhaps I should first remind you of the terminology here first. An AE value is a measure of how much a feature is overbet or underbet by the public. If the AE value is less than 1 for say all jockeys under 8st 5lbs then the public over bet these jockeys. If the AE value is greater than 1 then the public underbet these jockeys. To find the AE values you need to count the number of wins for the jockeys and then also the expected number of wins based on the odds. The last bit is the bone of contention. Should one first alter the odds so as to take out the overround. Instinctively I feel I should as this is more likely to reflect the true chances of the horse given the accuracy of the market generally.
However Phil made a very good point in that the overround adjustment is not really applicable to the same degree across all odds. The favourite long shot bias sugggest’s that we should adjust more for long shots than favourites. I must admit I tended to agree and as I was not doing this I expected to see some improvement if I left out the overround adjustment as Phil does on his AE ratings. I was surprised therefore to find that ignoring the overround actually made things worse. The performance of my top rated horses dropped significantly and as a result I have left things as they stand. I will continue therefore to run with AE’s based on calculating the overround for the race and then applying this to individual SP’s to get the running totals needed to calculate the final AE values.
If anyone has experience similar to the above I would be glad to hear your comments.