Sportstat gathered data frpm about 10 services including TG and Ragozin to do just such a study, after which Jim Bayle decided it took to much time away from his handicapping. He was going to run a whole bunch of mechanical studies-- best last figure, best top in last 5 starts etc., with no handicapping involved, testing win %, ROI, etc., and doing it for individual circuits, dirt, grass, etc.
There is no such thing as proof about something like this, just evidence. The study I liked best, which was my idea, was to take the last 3 figures for each horse (adjusted for todays weight), throw out the worst one, and average the other 2. This would give a pretty good idea who was best on recent form, which is about all you can ask out of a mechanical study.
