Track variant

[cosmicway]

I think it's difficult to find a universal method to compute track variants of horse races. It's central to building a good system plus, if you are an owner, to judge how your horse will perform against new opposition.

I know the story with the three youngsters running on the beach and the girl winning because she runs on the boardwalk and the other two run on the sand. But that's only how the story of the track variant begins.

I went on from that to construct my own as differences from the class par.
I 'm identifying the classes from the conditions declarations as described in the old cards -as best as that can be done- and proceed to calculate the difference of the winner's time from the mean par time.

But I 'm not satisfied.
First of all I use a logistic criterion.
There exists a universal criterion whenever you try to create/use software that predicts sports events of any type (horses-football-golf-discus ...). See here:

http://predictionscience.freeforums....ethods-t4.html

My track variant falls short.
It's the method I found in yahoo answers, ehow but falls short - compared to what has been achieved by others, who however miss out on other things (such as change of distance, track to track shipers).

So I 'm trying to figure out what to do.