I am wondering why so many horses do not have a JPR class number over "10".
All of the compound numbers in the program are created using a similar sequence of steps:
1. Individual factors are selected (or rejected) for inclusion in a factor mix based on win rate and roi observed in Data Window R&D using large samples.
2. Individual impact values and factor weights are determined for those factors that are going to be included in the factor mix. The factor weights themselves are based on observations made during Data Window R&D.
3. Algorithm Time... The weights are then multiplied by the impact values to produce a set of weighted impact values.
4. The weighted impact values are then multiplied together to produce a single "number" for every horse.
5. Each "number" is then run through a smoothing algorithm so that the number for each horse exists on a scale that makes sense to us human beings.
6. The numbers for each horse are then run through further algorithms that rank and gap the horses in each field.
JPRClass was developed in such a way that I purposely skipped step 5 from the above sequence. I wanted there to be at least one power rating in the program where the "number" itself represented the product of the weighted impact values.
So when you see one horse with JPRClass of 3 and another with JPRClass of 250 in the same race what you are really seeing is this:
IV1 x IV2 x IV3 x IV4... etc.
or
1.2754 x 1.8637 x .8976 x .7524... etc.
where etc. can be carried out for several hundred different individual weighted impact values.