Dosage: Pedigree & Performance
Main Menu         Search

The Relationship Between Racing Type and Breeding Type
Among Contemporary Sires

One of the common counter arguments to Dosage as a means of evaluating the distance potential of a Thoroughbred race horse is the racing type of its sire.  The argument usually revolves around the belief that if the sire had been, say, a classic winner or a BC Sprint winner, his runners should be capable of winning at classic distances or in sprints, accordingly.  If it truly was that easy, breeders would be breeding one legitimate classic contender after another.  Reality confirms, however, that this is not the case.  If breeding to classic-staying sires correlated strongly with the classic distance ability of their get I doubt we would see 90% of the field in the Kentucky Derby backing up in the stretch like Michael Jackson doing his moon dance.  What we might expect is a herd of horses barreling their way down the lane full of reserve energy and differentiated only by class and form rather than distance limitations.  We all know, of course, that's not the way it works.

The fact is that although there is a generic relationship between Thoroughbred racing type and Thoroughbred breeding type, it is a modest correlation and applies only to the Thoroughbred population at large and not necessarily to the individuals that comprise it.  The chart below displays the correlation between the average winning distance in furlongs (AWD) of a stallion both as a runner and as a sire.  The list includes prominent stakes-winning sires over the last 25 years that won at least five major North American open stakes races and also sired the winners of at least 25 of the same.

The correlation coefficient for the relationship is 0.64 which is certainly much better than random scatter but not exceptional.  Said another way, only about 41% (from R-squared in the chart) of the variation in the sires' breeding AWD is explained by the variation in their AWD as stakes winners. The other 59% is explained by other factors.  Could one of those other factors be the prepotent aptitudinal influences inherited from close up ancestors?  Dosage would say so.  I think it is fair to conclude, for prominent sires which had successful, high-end racing careers, that as a group, the shorter distance stakes winners generally tend to get shorter distance runners and the longer distance stakes winners tend to get longer distance runners.  The problem is that when creating a foal, the breeder is not breeding to a large, representative collection of sires, but just one.  Whether or not that one fits the general population profile is the issue.  Furthermore, if you include all sires, not just this select group of elite sires, the correlation coefficient between their AWD as a runner and their AWD as a sire drops to an unimpressive 0.33.  In other words, among all Thoroughbred sires the relationship between their AWD as a runner and as a sire is tenuous.  We can carry the analysis one step further and evaluate the relationship between a sire's maximum winning distance as a runner and his AWD as a sire of runners.  For all sires, the correlation coefficient is only 0.32, meaning that there is a marginal relationship between the longest distance at which sires were able to compete successfully and the AWD of their get.  Again, in individual cases the correlation may be outstanding, but those cases are atypical.  Finally, of all 890 sires of open stakes winners in our database, 62% have an AWD as a sire at least an eighth of a mile longer or shorter than their own maximum winning distance as a runner.  Clearly, projecting the winning distance of a racing prospect based on its sire having won the BC Classic or the BC Sprint is far more complicated than most people realize or will admit.  The projection becomes even more problematic when you try to ascribe a distance influence from a more remote generation.  

The following table lists the prominent sires alluded to earlier and includes their AWD as a runner, their current AWD as a sire and the difference between the two where a positive difference means they are breeding longer than they ran and a negative difference means they are breeding shorter than they ran.  The list is sorted from greatest positive difference to greatest negative difference.

There are several significant things to consider in these data which include 332 stakes wins by the sires as runners and 2,139 stakes wins by their progeny.  First, the group breeds shorter than it ran by an average of 0.25 furlongs.  This is most likely the result of a high concentration of mares in the breeding population bred for speed, a common situation in the modern era.  Second, 65% of the sires get progeny whose AWD differs from their own as runners by at least one-half furlong, some longer but the large majority shorter.  These are highlighted in gray.  Fully 28% have an AWD as a sire that differs from their AWD as a runner by at least a furlong.  In racing, a furlong difference in performance is what separates a sprinter from a miler, a miler from a middle distance horse and a middle distance horse from a classic stayer.  Third, based on the data, our expectations for sires often can be turned upside down.  Note the respective figures for the two stallions at the bottom, Lost Code and Marquetry.  The latter was a Grade 1 winner at the classic distance of ten furlongs and won four other graded stakes at nine furlongs, two at the Grade 1 level.  The former won eight graded stakes at nine furlongs, including two Grade 1s.  Yet between them they have sired the winners of 70 major North American open stakes races of which only 11% were beyond 8 1/2 furlongs and a stunning 71% were below a mile.  At the other end we find Distorted Humor, a sprinter/miler which won only one major race as far as a mile, all the rest being at sprint distances.  As a sire, 17% of his progeny stakes wins have been at classic distances and 74% have been at a mile or longer.  We also find Chief's Crown, a juvenile champion which came up short in all three Triple Crown races as the favorite in each.  Although he did win two Grade 1 races at a mile and a quarter after the Triple Crown, he was probably better at nine furlongs.  But even with his ability to win over a route of ground, no one could have imagined that as a sire of North American runners an astounding 40% of their stakes wins would be at classic distances or that a meager 5% would be in sprints.  These data don't even address his European runners that included Epsom Derby winner Erhaab.  And they don't reflect his influence as the paternal grandsire of Sinndar, the only horse in history to win the Epsom Derby, the Irish Derby and the Prix de l'Arc de Triomphe, all at 12 furlongs, in the same year.

This turnabout in distance is not unusual and we have often alluded to other prominent examples such as Slewpy, a classic-distance-winning son of a Triple Crown winner which became a premier sire of sprinters.  Sixty-six percent of Slewpy's progeny stakes wins were in sprints including a winner of the Breeders' Cup Sprint (G1).  Then there is Kingmambo, a multiple Group 1-winning sprinter/miler which has become one of the the world's great distance sires.  An unqualified international success, in North America alone his runners have won 22% of their stakes races at a classic distance.

The data suggest that many sires actually do breed on as they ran, with sprinters getting sprinters and routers getting routers.  On the other hand, the data also suggest that many others, in fact most, do not fit the expected pattern.  In some cases the difference between racing type and breeding type is dramatic and almost always unanticipated.  How could Artax become a sprint champion when his sire, his sire's sire and both second-generation sires of his dam (that is four of the seven sires in the first three generations) won Grade 1 level races from ten to twelve furlongs?  How could Smarty Jones, whose sire never won a major race beyond a mile and whose broodmare sire was a sprint champion win two American classics?

Therein lies the essence of the argument about the usefulness of chefs-de-race vs non-chefs-de-race is assessing distance potential.  What distinguishes chefs-de-race from other sires is their demonstrated predictability for passing on a generally uniform aptitudinal type to their descendants.  That type may or may not be similar to the type they expressed on the track.  It will depend greatly on the prepotent aptitudinal influences they carry forward from chefs-de-race close up in their own pedigree.  In the absence of aptitudinal prepotence one is essentially speculating or, more bluntly, guessing.  Sometimes the guess can be right.  There is at least an equal probability that it won't be.

Of course, the guessing approach is the historically traditional one, although no one thought of it as guessing because no one took the time to "do the math".  For years, breeders (and handicappers as well) have based their assessment of whether a future foal or today's runner will stay a given distance on the performance traits of their ancestors.  What we have learned over the years, however, is that by using the "guess" approach, very few breeders have been able to consistently breed legitimate classic horses and that even fewer handicappers have made a profit simply using pedigree analysis.  I think the situation is best summed up by paraphrasing Dr. Franco Varola, the originator of the aptitudinal approach to Dosage.  Varola made the point very clear when he said that the aptitudinal type of a sire is determined by what he passes on to his foals in the breeding shed, not by what he expressed as a racing stallion.  That remains today one of the single most important principles of contemporary Dosage methodology.