The idea is once you have completed your ratings, you then test them over, a number of series of races (this could be hundreds or more) you then would find the strike rate based on highest to lowest ranked runner according to your ratings, you can also isolate all runners according to your pricing (in percentile groups) an test your margin of error.
In your example your if your Top rater had a 36% strike rate followed by 18.9 for the 2nd ranked an 13.5 for the third ranked an 8.1 for the fourth ranked, your could compute your combined edge on this race and if it satisfied your min edge % criteria you would then either hedge &/or dutch all your runners.



[ This Message was edited by: woof43 on 2004-10-25 19:29 ]