I looked at the TG Derby archives to see how the various preps have help up in terms of the ability of their entrants to reproduce their final prep figure in the Derby. I probably made some errors in data entry, but I\'m looking at medians, so an error here or there shouldn\'t have a significant impact.
The median Derby runner ran went backwards 2 TG points from his final prep figure to his Derby figure. For the individual preps, the median backward move is as follows:
Wood - 0.4
Blue Grass - 1.1
Santa Anita Derby - 2.5
Arkansas Derby - 3.0
Lexington - 3.5
There are obviously sample size issues here. In particular, based on plotting the data, I suspect the Wood median would get closer to the Blue Grass with an increase in sample size.