Dosage and Claiming Horses

Virtually all of the published Dosage data are derived from North American open stakes winners.  The emphasis on stakes winners is intentional, and with good reason.  Stakes winners represent the highest level of Thoroughbred performance.  Horses competing in stakes races are generally in better physical condition than those competing at lower levels.  If physical problems do exist, the problems are likely better managed.  Horses competing in stakes races are more consistent.  If they stay healthy they can be expected to put in a competitive effort on a continual basis.  Horses competing in stakes races will usually be suited to the particular race conditions.  With larger purses on the line, and with a limited number of races in a horse's career, the connections of stakes horses tend to make management decisions that optimize opportunity.  This isn't always the case at the lower end of the class structure.  Cheaper horses are often raced into shape and their native talents easily can be obscured by chronic injury.  Their form cycles are more erratic than those of their more gifted peers.  Therefore, the intentions surrounding these runners are not always apparent.  With stakes horses, on the other hand, there is an excellent probability that they belong in their race.  By the time they are mature they will have found their best distance profile and most competitive racing class.  They are in to win, or at least get a part of the purse.  The result of superior talent that is well managed and purposefully intended is that the outcome of races involving those animals is a better reflection of their innate qualities.  In other words, it is a reasonable assumption that stakes horses more accurately express their genetic potential than do horses running in claiming races.  A secondary result is that the data obtained from their races is more reliable, and reliable data is absolutely critical if the objective is to develop a better appreciation of the relationship between pedigree and track performance.  Nevertheless, it is instructive to see how well the Dosage model applies to the lower end of the racing spectrum.

Fortunately, such information is available through the efforts of Mr. John Denbleyker and his student research thesis in the Sports Management Program at St. Cloud State University in St. Cloud, Minnesota.  Mr. Denbleyker investigated over 1500 claiming races in the years 2000 and 2001 at fourteen tracks:  Arlington Park, Belmont Park, Calder, Churchill Downs, Del Mar, Ellis Park, Gulfsteam Park, Hollywood Park, Keeneland, Lone Star Park, Pimlico, Prairie Meadows, Santa Anita, and Saratoga.  He calculated the Dosage figures for each of the winners and arranged the data according to racing category.  His results, reproduced with his permission, are presented below.  Included are the analogous data for just under 2000 open stakes races over the same time frame, allowing for a direct comparison. 

Data from the first table are shown graphically in the following charts.  The first plots the average DI for the winning claiming horses and the stakes winners against the average winning distance.  The second does the same for the average CD.  Again, the Dosage figure versus distance graph is a visual display of the fundamental Dosage model.

The graphical output of the data is unequivocal.  The general Dosage model applies to runners at the lower end of the class ladder as it does to stakes winners.  Also as expected, for reasons outlined above, we find a better fit of the data for the stakes winners.  This observation is confirmed by their higher R-squared values, both in the DI and the CD plot.