Classhandicapper:

As someone who used to make figs, it would all depend on how you base them. Track records? Pars?

I would always "get in trouble" with a race like the Belmont. I found out early that you can't use, say, Secretariat's 2:24, as it was an abberation. On a "par" track, a number closer to a 2:27 works much better. To better quantify this, I feel a 3 point regression from negative numbers (Smarty) is not as bad as drop from a 7 to a 10.

I'll keep saying this: w/o Birdstone, Smarty wins by 8 in what would visually appear to be a romp, because his rate of deceleration is not as severe as the rest of the field.