podonne
10-05-2008, 10:35 PM
In Appendix A of "Winning Thoroughbred Strategies" (Mitchell, Dick 1989) The author provides this formula for determining the minimum number of races to test a system on, based on the win percentage, average mutuel, and the minimum error range. The formula is as follows:
N = W * L * (Z/E)^2
Where W is the win %, L = 1-W, Z is a constant based on a z-score for your desired confidence level (for 95% use 1.96) and E is the maximum allowable error (this is a complicated formula using the average mutuel I won't include, for copyright's sake, I only show the formula so you will know the formula I am referring to)
I wonder if anyone has taken the time to apply this formula (or like it) to determining the minimum number of "plays" or "entries", rather than races. I though of just multiplying the number of races by the average field size, but that seems to simple.
I would also be interesting in similar formulas to accomplish this. This formula is interesting in particular because is needs the average mutuel, i.e., the higher the win %, OR the higher the average mutuel, the lower N will be. I like formulas that require some component other then just win %, either expected winners (summation of each win probability) or average mutuel.
Thanks,
Phil
N = W * L * (Z/E)^2
Where W is the win %, L = 1-W, Z is a constant based on a z-score for your desired confidence level (for 95% use 1.96) and E is the maximum allowable error (this is a complicated formula using the average mutuel I won't include, for copyright's sake, I only show the formula so you will know the formula I am referring to)
I wonder if anyone has taken the time to apply this formula (or like it) to determining the minimum number of "plays" or "entries", rather than races. I though of just multiplying the number of races by the average field size, but that seems to simple.
I would also be interesting in similar formulas to accomplish this. This formula is interesting in particular because is needs the average mutuel, i.e., the higher the win %, OR the higher the average mutuel, the lower N will be. I like formulas that require some component other then just win %, either expected winners (summation of each win probability) or average mutuel.
Thanks,
Phil