I can\'t swear to the accuracy of my hand-entered data, but for horses that have run at least two points worse than their top in their last prep (a sample of 70), I get:
New top: 4.3%
Pair: 12.9%
Off: 32.8%
X: 50%
The three new tops were Liaison (2 pts), Imawildandcrazyguy (1.75), and Ten Cents A Shine (1.5).
Note that I\'m not necessarily looking at bounces (poor effort following a good one), just how their last prep compared with their previous top, whenever earned. For example, Classic Empire would fall in this group this year, even though his last race was better than the one that preceded it.