While this may prove to be interesting, these particular examples are not too compelling (I must say I didn't see them before the races).

Hoffmans -- The small sample is the least of it. Adoration could have run a new top (1) and still finished behind five of her competitors in the race, right? If you could say "61% of his horses ran six point (average) new tops over the last 90 days" then you've got something.

Mandella -- Ditto. How big are the new tops? Nobody needed this stat to use Pleasantly Perfect after he won three prior races on the card. He would have been 25-1 at least if it wasn't for everybody piling on the bandwagon. Also another who could have run a new top (negative 1-1/2?) and finished behind at least three or four horses in the race. You would need a stat like "40% wins with horses running a new top second off a layoff in a race where the four or five other likely candidates don't fire or burn up dueling for the lead."

Frankel -- You said it. The "5% X" stat was overshadowed by the "Frankel sucks in the Breeders Cup" stat that everybody had ahead of time and still bet his horses like crazy right up to the Classic.

The main thing I would like to see in the trainer stats is more "meet specific" or "track specific" info. If the TG-based stuff could reflect this kind of data I would be a lot more interested. And obviously quantifying the "new top" thing would be good. HP