Trainer Stats: We use a
Composite Approach. That is, there are like 150 or so groups of trainer angles. When I say, "Group," I mean like "distance."
"Distance" is expressed in multiple ways. For example:
1. Distance= sprint or route
2. Distance= mini, dash, sprint, route, marathon.
3. Distance= Surface & Distance, such as Dirt-sprint, dirt-route, turf-sprint, turf-route.
Thus there are multiple looks at the distance angle.
Overall, one could (logically) divide trainer stats into 3 categories:
1. Control Factors - Those angles that the trainer has control over such as class change, distance switch, days off, etc.
2. Abilities - Those angles that are typical of the trainer's characteristics such as, age, sex (eg. some trainers are very good with 2yr old fillies but not so great with 2yr old colts), class levels and race types.
3. Performance - Those angles that are representative of the horses relative ranking in today's race. For example, if the horse is ranked 3rd for speed rating in the last race, how does he do with such horses? What if he ranks in the rear half of the field?
Same thing for best-of-last-2, average-of-last-3, or earnings box-class.
By far, the most important stats are the performance stats because nobody - at least nobody I am aware of - has such statistics.
Each user has the option of turning on or off whichever factors they wish to consider. In addition, there are a dozen or so columns for each statistic, such as Impact Value, Pool Impact Value, $Net, etc. and a bunch more. The same stats are available for place and show as well.
Each user has the capability of "filtering" the factors that qualify. For example, they might say that only factors where they had a 30% hit rate (or higher) should be considered, or only factors where the horse had at least 4 wins. (Ed Bain called this 4+30.)
Also, each user can create their own "handicapping objects" from the statistics derived from the qualifying factors.
All trainer stats are "circuit-specific." Thus, stats from SAR, BEL and AQU are co-mingled, while PARX stands by itself. IMHO, national statistics have very little value.
Perhaps the very best example of this has historically been Jerry Hollendorfer. For years, Jerry was a 35-40% trainer in NoCal, but when he shipped to SoCal he was a 13% trainer. If his stats are computed nationally, he will be way overrated in SoCal and actually a little underrated in NoCal (by virtue of his SoCal races drawing him downward).
BTW, we take all of the stats, for all of the qualified columns and tally them together. Works like a charm but do not try to do this without a software program. LOL
Oh, and we do EVERY trainer, in EVERY race, at EVERY track. Their tendencies cannot be hidden from us.
I believe this video will show a little of how it all works.
http://www.practicalhandicapping.com...inerStats.html