As we get ready for the start of the Northern Hemisphere yearling sales series, one of the more pertinent questions that we are often asked at TrueNicks is how important is the racing class of the pedigree at hand in determining the success of a mating. While the TrueNicks rating of a yearling can be the same, the chances of a stakes winner occurring from one mating over another can vary when we look at large populations of horses.
In order to start to look at answering this question, we needed to have a baseline of measurement of racing class. One such measure is the Class Performance Index (CPI). Developed by the Jockey Club Information Systems, CPI uses prize money—which while not optimally distributed is at least some measure of racing class—to create an index ranking which is calculated for each year and based on:
- Horses of the same sex (i.e. only males compared to males and females compared to females)
- Horses of the same age
- Horses racing the same year(s)
- Horses racing in the same countries
All of the criteria must match for the comparison to take place. To calculate a horse's CPI, the horse's average earnings per start is compared to the average of the average earnings per start of all like horses. That is, if a horse's CPI is 5.0, then his average earnings per start is five times greater than the average of all horses of the same sex, age, and country, racing in the same year. It is as good a measure using prize money that is available and allows us to compare one generation of horses against another across a population.
Now that we have established a measure, we need to establish a metric of performance that we can use to compare from. To do this we looked at all the yearlings that were offered at the three major yearling sales (Fasig-Tipton July, Saratoga, and Keeneland September) from 2008 to 2010 and pulled their subsequent CPI's. From the 14,864 horses that were sold at these sales, we established that the CPI isn't an exact nor linear measure of racing class in that you have a horse like I'll Have Another with a CPI of 104.62 and Game on Dude at 71.57, so over 30 points difference, and a horse like Centralintelligence, a grade I winner, has the same CPI (7.50) as Buffum, a grade III winner.
That said we can make some reasonable assumptions and groupings from the data
- Group 1 (494 horses or 3.3%) - these horses have a CPI greater than 5.0. These are your "elite horses" in that they are good money earners. There are plenty of non-stakes winners in this group but they are high end allowance level horses that earn significant prize money.
- Group 2 (794/5.3%) - horses with a CPI between 2.75 and 4.99. These are your "wash" horses. There are a few horses that return more than they cost to buy and race but in general this group of horses that had a career that resulted in a wash in terms of cost/return. There are plenty of lower level stakes horses in this group but the bulk are just winners.
(11627/78.2%) - horses with a CPI between 0.00 and 2.74. These horse made it to the racetrack but are losing propositions in terms of cost/return and are in general limited racehorses with lots of claiming level horses
- Group 4 (1949/13.1%) - these horses never made it to the track.
Interestingly the 3.3% Elite horses is just about breed average (3.1%) for the Grade/Group stakes winners out of the top 1% of all stallions, so the CPI cutoff of 5.0 looks to be a reasonably solid metric. It's also interesting to note that 87% of the yearlings sold at these sales made it to the races, which is significantly higher than breed average (72.3%), but you would expect that coming from these three major sales where the subsequent opportunity (trainers, vets, etc.) is a lot higher than the breed average.
So now that we have established some baselines of measurement we had to take out the influence of sire power on the results. To do this we took a separate population of elite stallions and looked at the CPI's generated by their offspring and the CPI's of the mares that these foals were bred out of. The stallions that we selected were:
These eight stallions are somewhat comparable in terms of their performances at stud, so we can assume that the effect of "sire power" on their collective offspring is accounted for. While it is hard for any stallion to come up with the raw SW/foals ratio generated by Northern Dancer, he operated in an era of smaller books with larger books flattening both SW/foals ratios and also the AEI of the stallions.There are 8185 horses by these stallions that generated CPI's from the 10960 foals born which is about 80%. It is interesting to compare the distribution of CPI's in the Groups with the group of all yearlings that we detailed above.
Again there are some noteworthy observations. Firstly the percentage of unraced horses by these sires is a lot higher than the yearling population. I'd like to call this the "I'm probably useless but well bred so don't expose me as useless" effect.If a well bred horse by one of these elite sires is known to be slow, they are not tried on the track, rather sent to the breeding shed to breed. That is probably not the best description of the total effect because there are probably mares in that group that were "hard luck" horses, that never got to the track because they hurt themselves but they had talent (i.e. they were too fast for their own good). The other observation from the data is that these eight elite sires got almost four times as many elite horses (those generating a CPI of 5.0 or greater) than a general population of stallions.
Obviously stallions such as these eight get some skew in the quality of the mares that they breed. 27.7% of the mares that they bred had a CPI themselves greater than 5.0 (thus in Group 1), however 42% of the mares that they were bred to were unraced and didn't generate a CPI at all. So here is where we can use the "relative risk" calculation to begin to answer the question "All other things being equal, how much does the racing class of the mare (as measured by CPI) increase/decrease the chance of the foal becoming a superior runner (that being a horse that generates a CPI of 5.00 or better)?"
|Dam Group||CPI Range||Horse Group 2-4||Horse Group 1||RR|
So what does the above table tell us? Firstly, if you are buying a yearling, or selecting a mare to breed from, all other things being equal if the dam of the yearling has a CPI of 5.00 or greater you have a 42% better chance of buying/producing an elite horse (one that itself generates a 5.00 CPI) than if you buy or breed out of a mare that doesn't have a CPI greater than 5.00. That is a lot of advantage for the yearling buyer and it is also a good bit of knowledge for the broodmare owner given that we said earlier a CPI of 5.00 or better doesn't necessarily mean that you are a stakes winner (something to consider when buying a mare later in the year).
The other observation is that the foals bred out of unraced mares (Group 4) have a significant but small (11%) decreased chance of becoming a superior runner, which is noteworthy when compared to the Group 3 mares that exposed themselves on the track as limited runners. This is probably generated by an effect that I mentioned earlier in terms of "hard luck" mares. If you have an unraced mare and were thinking about sending her to one of these elite eight stallions, you'd like to think that breeders knew a little bit more about her ability as a runner even though she was unraced, so the percentage of "hard luck" mares in the unraced group is probably higher than normal. The other observation is that those mares that do expose themselves on the track as limited runners (Group 3) represent a significant disadvantage in terms of producing a superior runner at stud. They can still get a good horse (Jude, the dam of Yesterday and Quarter Moon, was as slow as they come on the track) but as a group they represent a considerable disadvantage to the breeder and yearling buyer.