Perry Martin: Observations of a Breeder Pt. 3

Previously I discussed our first racehorse, Searchforthetruth, and how I liked his chances in an upcoming stakes race at a Northern California fair. Anyone who looked up his race record might be puzzled as to why, with a 5-5-5 record looking just at the 23 starts for Blinkers On, I would think he had a good shot to win his first stakes race. Good question. The answer lies in breaking down the overall data into sub-sets. 

 

Have you heard the phase, "Horses for courses?" If you look at "Search"'s 14 races run at Golden Gate Fields and Bay Meadows, there is only one win (7%). If you look at his nine races run at other tracks in California there are four wins (44%). He was a completely different horse on those racing surfaces. One of these wins was on the dirt at Solano, going six furlongs in 1:09.03! He won for fun that day pulling away from the field, and I believe he would have done well on any of the Northern California fair dirt tracks. I'm not trying to provide a handicapping lesson, only to demonstrate the importance of defining, separating, sifting, and prioritizing data subsets for inclusion into an analysis process. Now let's apply the process to stallion selection.

 

I believe I used both of the available computer-based nicking programs to evaluate the Lucky Pulpit  /Love the Chase match. At that time one gave me a "C" while the other gave me a "C+". Anyone out there interested in breeding an average horse? Me either. Why then would I select that match? Easy, I didn't believe the computers. By the way, I just ran the same match on one of the programs and got an "A." It's getting better over time. To anyone not familiar with nicking, the central idea is that all bloodlines are not the same and that some specific combinations do better in producing desirable offspring than other combinations. What is better? Although the software algorithms are proprietary, they all look at some combination of percentage of starters, percentage of winners, money won, and number of stakes winners. That's easy enough, but then they use proprietary biasing protocols, giving more or less weight based on generation (closer carries more weight), sub-group populations (smaller groups less weight), etc. When evaluating the results, one must consider other biases that are not addressed in the software. One example that applied to Lucky Pulpit at the time was that new sires usually do not receive the best mares.

 

Before concentrating on what sire I selected for Love the Chase, it's probably best to take a step back and look at the process I used to find him. Again, you have to look at the mare's bloodline first. I made a decision that I would only look at "clusters." A cluster is the result of breeders' human nature. It occurs when one breeder produces a very good horse. Other breeders take notice and duplicate the cross, hoping to also get a good horse. The result is a fairly large population of a specific bloodline cross that is statistically significant. Being statistically significant simply means the results are more believable and not a fluke. As it takes multiple breeding seasons and years of racing results to develop a cluster, it is necessary to skip a generation in the pedigree for the analysis. For this reason I looked at the performance of Mr. Prospector mares against the major sire lines.

 

To make a long story short, one of the most intriguing crosses I found for Mr. Prospector mares was A. P. Indy. A. P. Indy through Seattle Slew brings in two more lines back to La Troienne. There were 71 different Mr. Prospector mares bred to A. P. Indy resulting in 127 foals. Of these 103 (81%) started, 84 (66%) were winners, 22 (17%) were stakes winners. Better than one in six of all foals were stakes winners. Among all starters, 21% were stakes winners--better than one in five! That's what I'm looking for. Of course, the experts will say don't look that far back and expect similar results; the genetic factors of importance will be diluted. I don't disagree with that statement. However, if you offset that influence by concentrating factors using FFI aren't you back to average? Anyway, I'm looking for better results. 

 

A simple, common breeding pattern Edward Stanley followed used inbreeding to concentrate preferred traits then outcrossing to correct deficiencies. Note that Pulpit was an A. P. Indy/Mr. Prospector mare cross. Pulpit now has 30 sons at stud around the world. This is considered one of the major Thoroughbred bloodlines in the world with the current top American stallion being Tapit. Yet the line has its detractors. Many breeders in America point to a fragility of the line because there is too much inbreeding. Too many of the stars coming out of this bloodline are retired quickly due to injury. The poor showing of A. P. Indy offspring in Europe has breeders there saying the line has no talent on turf. Wanting the high talent benefit of the A. P. Indy/Mr. Prospector mare cross without the deficiencies meant looking for a son of Pulpit with a beneficial outcross.

 

Lucky Pulpit being out of a Cozzene daughter was one such candidate. His standing in California so that we could take advantage of the Cal-bred benefits made it a no-brainer. Cozzene was sturdy and provided yet another line with turf influence. Our statistically insignificant one-horse sample indicates any deficiencies may have been corrected. I'll withhold judgment for a few more years, yet I'm encouraged. I know one race on the turf will not convince anyone. We have targeted several high-profile turf races for Chrome this year. Hopefully, he'll show something that will make European breeders come back to this bloodline.

 

Modern breeding theory in America has been highly influenced by Joe Estes, the former editor of The Blood-Horse. Estes understood statistics and applied them successfully to breeding. Before Estes, analysis shows that major successful breeders were obtaining one stakes winner from every 48 foals. Breeders using Estes' methods were able to obtain one stakes winner from every eight foals. That's pretty good. But not everyone really understands statistics. Over time Estes' methods became oversimplified and sometimes misapplied. I have studied these methods; however, it is difficult for me to boil them down to one article. Most proponents pare it down to just one sentence: Breed the best to the best and hope for the best, which is not very helpful. What is the best? Racing performance is an indicator of the presence of beneficial genetic factors. Would A. P. Indy be an inferior stallion had he never raced? I think not. How those factors combine across bloodlines is a different problem.

 

The highest stud fee does not make a stallion the best for your mare. The people who own that stallion will say differently because that's how they make their money. A high stud fee implies the best. A low stud fee implies poor breeding to those ignorant of what's behind good breeding. Again, you can be a successful breeder through luck, but I believe it's much better to be successful through understanding chance. It means study, understanding, and extra work, but it's worth it.

Both our understanding of genetics and the development of testing technology are expanding rapidly. We will soon be at a point where tests will determine optimal breeding combinations. It is comforting to note that once those selection methods are underway and have advanced the breed through several generations, horses will conquer the Earth and run the show. The racing industry will be saved—maybe.

 

READ: Perry Martin: Observations of a Breeder Pt. 1