The "claiming par" conversation took place a couple of years ago. In answer to a question, Friedman said that the reason their figures could be used to compare horses from different generations was that they had a solid base, par levels based on the claimers, which stayed the same year after year. I responded under an alias, pointing out that keeping the par levels was an assumption (one of a large number that operation makes, but that's another story) with no basis in fact. If performance of the breed improves over time (as all other athletes have), the claimers would improve as well, and artificially anchoring the pars will create figures that CAN'T be compared for different generations, by their very nature.

Pars are a very good way to set up your data base to start. But they have drawbacks in use over time, and not just the above. First of all, they are based on a relatively small sampling-- the winners (or some variation involving more horses) of just SOME of the races (usually older horse claimers, since the winning figure in those races-- as opposed to the horses themselves-- are the least variable), compared to an AVERAGE race of that type (par for a 10 claimer, for example). As soon as you have a relatively solid data base, you are much better off losing the pars and going with the "projection method", which involves looking at ALL the horses in ALL the races (obviously paying more attention to the more solid ones), and comparing their figure performances not to an average 10 claimer, for example, but to the figures the horses in THAT 10 claimer have run before. This gives you a lot more "data points" (to use a term someone else used here once) to work with, and frees you up from artificial anchors.





TGJB