The obvious problem with using pars to check what you call biases (as opposed to dead rails etc.) is the self fulfilling aspect-- if you check figures against rigid pars, and the horses themselves are evolving, it won't be reflected in your figures.

In general, I don't base a variant on lightly raced horses, other than fine tuning one that is already established for the race by surrounding races with a consistent pattern. On days where the track is changing speed, or there is reason to believe such a race should not be done with the other races, I leave a box instead of a figure (if you go back in the archives you will see an extended brouhaha about this subject focusing on Chilukki's debut figure, I think the debate took place early this year). I don't like to guess, in part because future decisions will be based on those figures.

Again, pars deal for the most part with just the winners of the races, where what we do deals with all the horses. Part of this is not scientific, and it goes like this-- after doing this a long time, you get a feel (even if you can't articulate it) for roughly what percentage of horses should be running new tops, pairing tops, etc., keeping in mind the ages and number of starts of the horses in question. Related discussions about this have come up here before-- one with David Patent about the Schaeffer run at Pimlico on Preakness day 02, when Ragozin gave a whole field of older stake horses figures at least 5 points off their tops, or something like that. Another was this year's Derby, where Ragozin had NONE of the first four finishers even matching their tops, due in part to blowing the Wood figure, which I pointed out before the Derby. That is flat out ridiculous.



TGJB