Super Saver: a Minority Report
by Roger Lyons
In a recent post on the breeding behind Super Saver, Sid Fernando sets in relief three differing approaches to identifying pedigree factors that account for that horse’s ability to win the Kentucky Derby, including my own. Nobody does the historical color as well as Frank Mitchell, but, clearly, Alan Porter’s analysis, as endorsed by Andrew Caulfield in his TDN column of May 4, is at odds with my account in regard to the factors that it emphasizes.
I realize that Andrew’s analysis has the advantage of being the standard account. Its topical structure defines the prevailing pedigree analysis, including its center-stage placement of linebreeding as the main actor. Therefore, I offer my rebuttal as a minority report.
Linebreeding certainly does have a functional relation to performance, but I emphatically disagree with the presumption that its effects must be favorable, much less decisively so, just because it happens to be found in the ancestry of a Kentucky Derby winner. To take Super Saver’s performance as evidence of the benefits of linebreeding simply begs the question of its effects and, more broadly, of its functional relation to performance. Unless the standard account can establish a statistical relation between Super Saver’s linebreeding and his performance, then its analysis might as well be founded on a non sequitur, and I believe it is.
As Sid points out, Super Saver’s family began with “a simple nick between a La Troienne daughter (Baby League) and War Admiral.” The family continued to evolve on that same basis, each daughter having had her best production by the best stallion that afforded the best simple nick. The accumulating linebreeding, involving multiple strains of La Troienne, therefore, was a systematic consequence of that series of simple nicks. More broadly, linebreeding is a systematic consequence of the pedigree model of breeding. It’s not a formula for breeding great racehorses. Rather, it’s an inevitable consequence of the process.
So, is linebreeding a cause of superior performance or just an effect of pedigree breeding? The latter is evident from the genealogical record, but inferring a causal relation to performance requires populational evidence that not even the cheerleaders for linebreeding are able to provide.
Well, then, what can be inferred about it? We know what the effect of linebreeding is in a broad sense. Its purpose and effect is to establish a generational continuity between a new individual and its ancestry. The more linebreeding a pedigree has, the more likely is the individual representing that pedigree to express the traits conferred by ancestors to which it is linebred. But, if linebreeding is so effective at reproducing the traits that constitute a great racehorse, why is it that so many horses with linebreeding similar to that of Super Saver in distant generations are just no good?
I firmly believe that the prevailing pedigree analysis–the standard account of pedigree–misunderstands the way in which linebreeding is functionally related to performance. Let me explain by analogy.
A good melody consists of two fundamental elements that oppose one another at every musical level. In order to be recognizable as such, a melody must have continuity. It must have a certain repetitive rhythm. If sung, the words must rhyme. A refrain is very much a part of what we expect of a melody along this direction of its movement. However, the continuity of good melodies is subverted at every point and at every compositional level by the element of variation moving in the opposite direction. Each measure of a melody must be continuous with the last, but different from it. The lyrics rhyme by repeating the same sound, but enunciated in different words. The refrain interrupts and contrasts with the sequence of verses. A good melody arises from the tension between continuity and variation, the latter always playing a subversive role.
Breeding a good racehorse is just like that. Linebreeding mediates generational continuity. Its function is to specialize the new individual around qualities that are conferred by the ancestors to which it is linebred. At its best, it yields a physically coherent individual. However, the new individual must also be capable of a well-rounded performance. It must have the variety of typological possibility required by the prevailing conditions of racing. That’s the job of generational variation, which operates in opposition to linebreeding.
In the same way that musical variation subverts continuity in the making of a good melody, generational variation subverts the continuity established by linebreeding. This fundamental opposition between linebreeding and generational variation is what the advocates of linebreeding don’t get.
According to the numbers I have for Maria’s Mon as a sire (and other stallions as well), A.P. Indy, Supercharger’s sire, has just such a subversive relation to his own linebreeding to La Troienne. After all, Maria’s Mon hasn’t otherwise done that well with Buckpasser (10 superior runners from 156 mares through his 2007 crop, counting the three through A.P. Indy). If you take out the seven mares in descent of A.P. Indy, then Buckpasser’s strike rate falls to 7/149.
Nor has Maria’s Mon done all that well with mares in descent of Seattle Slew, another important source of La Troienne. If you take A.P. Indy out of Maria’s Mon’s strike rate of 5/46 with Seattle Slew, it drops to 2/39. In fact, all but one of those five superior runners were out of mares with Seattle Slew in tail-male descent. But, if you exclude A.P. Indy mares, the strike rate with Seattle Slew in tail-male descent is only 1/19. So, read very carefully how the standard account gets its numbers relating to the Maria’s Mon-Seattle Slew “nick,” including that restricted stakes winner thrown in for good measure, because there’s much at stake in it for the linebreeding hypothesis.
My numbers, by contrast, don’t suggest broadly favorable effects of linebreeing to La Troienne through the ancestors of A.P. Indy. What they suggest is that A.P. Indy’s influence very favorably disrupts effects of linebreeding that are otherwise not at all favorable to foals by Maria’s Mon. A.P. Indy provides a beneficial variation that combines with Maria’s Mon to yield a simple nick.
It’s quite possible, too, that Numbered Account, a key source of La Troienne also tends to subvert the La Troienne continuity. Maria’s Mon has a strike rate of 1/3 with daughters of Numbered Account through his 2007 crop, but her dam, Intriguing, otherwise has a strike rate of only 2/23. Maria’s Mon happens to work with Numbered Account, as far as can be determined, but not so much with her sire, Buckpasser, or with her dam.
As a matter of fact, the ancestors of Supercharger that have had the most favorable impact for Maria’s Mon through daughters have nothing whatever to do with La Troienne or with A.P. Indy. Maria’s Mon has a strike rate of 7/50 with daughters of Mr. Prospector, sire of Super Saver’s second dam. With daughters of Northern Dancer, sire of Super Saver’s third dam, Maria’s Mon has a strike rate of 5/34. These numbers, far from confirming that Super Saver’s performance is an effect of linebreeding, clearly suggest that it’s more likely an effect of generational variation. The influence of these important sires interdicts a linebreeding continuity that otherwise really hasn’t worked for Maria’s Mon.
In one pedigree after another, ranging across many different sires, the numbers say linebreeding is not the decisive factor. In fact, the numbers point to the ways in which atypical or variant strains subvert the effects of linebreeding. Such evidence trends toward the inference that Super Saver’s ability to win the Kentucky Derby is decisively affected, not by his linebreeding, but by the various directions in which his ancestry has deviated from its linebreeding to constitute a well-rounded individual–a horse with the speed, stamina, stoutness, and physical courage not only to withstand the rigors of training for that race, but also to win it.
We live in an era during which linebreeding has become ubiquitous in the population. All of the horses are linebred, and the functional relation of linebreeding to racing performance has already been taken too far. It’s too late to turn to the standard account for celebrations of linebreeding. In such an era, the successful breeder is the one who can identify useful variations with which to restore the residual aptitudes, the one who understands that linebreeding is the problem, not the solution.
Posted by Roger Lyons on Friday, May 7, 2010 at 12:21 pm.
Having it both ways?
Unfinished business left over from my last blog post really needs to be addressed. The point of that post was to show that Alan Porter uses deceptive language in characterizing his nick rating system’s access to information relating to the measurement of opportunity. Such tactics only go so far before colliding head-on with reality, and, when used so carelessly and with such abandon, they are usually part of a larger pattern of deception. That is, they are habitual.
This is evident especially in the hypocritical nature of Alan’s stated commitment to using opportunity as a measure of the effectiveness of sire-line crosses. The pattern of duplicity is pervasive.
As I sat down to compose this post, I decided I would just go to his website and use his latest post as an example, whatever it might be. I knew it wouldn’t matter because they’re all the same. It happened to be an entry about the More Than Ready-Meadowlake cross (“Halo Effect”), and it’s very typical of the stuff he does all the time. Everybody knows that Alan’s big thing is to tie up the loose ends of ancestors that share a close genetic relationship–in this case Halo (More Than Ready’s paternal grandsire) and Prince John (Meadowlake’s tail-male great-grandsire).
As everybody also knows, Alan and his partner Byron Rogers are principals in a nick rating service whose main selling point is that it uses opportunity against which to measure the effectiveness of sire-line crosses–this, despite the fact that Jack Werk, Sid Fernando, Bill Oppenheim, and I (and probably others I don’t know about) have said for years in print and online that opportunity at that level is meaningless.
The reason is that, when dealing with descendents of sire lines and broodmare sire lines, there’s no way to control for the quality of breeding stock. In that case, when you reduce opportunity to “the number of times the cross has been tried” (let’s call it N for short), you are bound to end up with a wild discrepency between the two. Never mind Alan’s naive, but calculated, use of the term “true opportunity.”
Given a commitment like that, it’s surprising that, in making his point about the Halo-Prince John effect as it relates to More Than Ready, Alan never mentions how many of More Than Ready’s mates actually had Prince John in their ancestries. This is all the more surprising in view of the fact that the question of opportunity has its greatest relevance when it relates to the mates of a particular stallion because, in that case, the stallion himself controls to a great extent for the quality of mares. Thus, in that situation N is much more representative of actual opportunity. Why does Alan abandon his commitment to opportunity in such circumstances?
In my small consulting business that’s exactly the kind of information I use all of the time. For example, my database for More Than Ready shows that through his 2006 crop (I add the latest three-year-old crop in the middle of the year) he’d sired foals out of 87 individual mares with Prince John in their ancestries (North America only), and some of those mares had produced multiple foals by him. To date, ONLY three of those mares have produced a stakes winner by More Than Ready.
By the test of statistical significance I use, that’s an abysmal strike rate, relative to More Than Ready’s overall record. Alan is simply wrong about the More Than Ready-Prince John connection. Well, what about the greater Halo-Prince John effect that’s really at stake in his post?
I checked the performance with Prince John by other sires with Halo in their ancestries. Street Cry is the only one whose strike rate with mares in some descent of Prince John was significantly higher than his overall record. Fusaichi Pegasus, Giant’s Causeway, Pine Bluff, Rahy, Saint Ballado, and Victory Gallop all had SW strike rates that only just warranted their opportunity. The rest of them for which I”ve kept figures–Devil’s Bag, Harlan’s Holiday, Silver Ghost, and Van Nistelrooy–had SW frequencies with Prince John that fall significantly below opportunity, relative to their overall quality as sires. Conclusion: there is no broad, performance-enhancing affinity between Halo and Prince John.
A proper consideration of opportunity would have prevented Alan from misleading his readers. Unfortunately, while insisting upon the need for a measure of opportunity when its use is dubious at best, he completely ignores it when it would have the most relevance. He constantly reminds us that he has access to vast database resources, which leaves only one explanation. His commitment to the question of opportunity varies, not with the matter of its relevance, but with what it is he’s selling at the time.
NOTE: The above references to SW strke rates for individual stallions with mares in any descent of Prince John are partially derived from source material obtained from The Jockey Club Information Services, and I thank my lucky stars for policies on the part of that organization enabling small businesses like mine to participate in the thoroughbred industry in a way that’s meaningful and cost-effective.
Posted by Roger Lyons on Tuesday, January 12, 2010 at 12:03 pm.
• Permalink • Comments Off on Having it both ways?