Ruth N. Bolton and Randall G. Chapman published “Searching for Positive Returns at the Track: A Multinominal Logit Model for Handicapping Horse Races” in Management Science, Vol. 32, No. 8, August 1986.
The study concluded that the most significant factor to look at was the “average speed rating” (over the last four races). Of much lesser significance, but still worth noting, was “winnings per race in the current year,” followed closely by the winning percentage of the jockey. The study also concluded that post position and a new racing distance were meaningless variables, yet most people still waste copious amounts of time worrying about such matters.
I have taken things to the extreme. I do not look at the past performance records of the entrants in a race, choosing instead to watch the tote board in search of betting anomalies. However, not everyone has the ability or the temperament necessary for such an endeavor.
It would seem logical then that the typical race goer should begin the handicapping of a race by averaging the speed ratings for all contenders in a race, then move on to winnings per race in the current year and the winning percentage of the jockeys.
(It should be pointed out that the speed ratings used in the above study were the traditional Daily Racing Form speed ratings, and not the worthless Beyer Speed Figures that now occupy space on a past performance record.)