Am I missing something here, or wouldn't you end up with a few groups all with the same price ? Going no further would be OK for 2 and 3yr. olds where a min. horse profile exists but beyond that age range the prices would become misleading.

I often use a similar method to get an initial [base] rating figure and then divide each runner's number into the total to obtain a probability as you both have done, but then I divide SR by 3 [add a 0 to SR] and adjust up or down according to good or bad stats [+10 for a win at the track or a -10 for no wins etc. as an example] and finally add this to the initial base rating and convert to price.

PS. I use a 1 to 10 method for my base rating without adding a 0 to the SR and use +1 or -1 for stats but I've worked it above to fit your methodology.

Try it out.