Does Pedigree Matter Any Longer in the Kentucky Derby?

Following the startling upset win by Giacomo in the 2005 Kentucky Derby at odds of greater than 50-1, some observers have speculated that pedigree is no longer as important as it once was in defining the classic American Thoroughbred.  After all, Giacomo's sire, Holy Bull, had previously sired only one major stakes winner on the dirt at a distance beyond nine furlongs.  That horse was Thunder Blitz, winner of Aqueduct's 2003 Grade 3 Queens County Handicap at 9½ furlongs and beaten in Monarchos' 2001 Kentucky Derby by almost nine lengths.  Adding to concern about Holy Bull as a classic sire was the short 7.65 furlong average winning distance of his progeny in North American open stakes races. 

The situation is complicated by the recent Derby wins of Smarty Jones in 2004, Funny Cide in 2003 and War Emblem in 2002.  They were all sired by young stallions, none of which had ever won a stakes race beyond a mile.  Elusive Quality, sire of Smarty Jones, and Distorted Humor, sire of Funny Cide, were noted for their blazing speed, the former setting a world record for a mile on the turf and the latter setting the seven-furlong track record at Churchill Downs.  War Emblem's sire, Our Emblem, could do no better than place in graded stakes races at seven and eight furlongs.  There is little to indicate from their racing careers that they would become classic sires early in their stud career.  Of course, the sire side of a pedigree is just one half of the equation.  A speed sire alone doesn't preclude the possibility of a classically-bred, aptitudinally-balanced pedigree.  Which brings us to Dosage.

Following a half century of Derby wins by horses with a Dosage Index (DI) of 4.00 or less, we have seen four winners exceeding that figure since 1991, including Giacomo.  Now for many people these events are proof that Dosage has lost its usefulness or never was useful in the first place.  That is an obvious conclusion for those who considered Dosage to be only a mysterious magic formula for handicapping the Kentucky Derby.  More than likely they are unaware that Dosage is, in fact, a pedigree classification system intended to monitor the dynamic changes in the relationship between pedigree type and track performance.  If pedigree type is truly evolving, Dosage will capture the nature of that evolutionary process very well.

Dosage methodology has enabled us to observe an "inflation" factor in the DI of Derby winners for many years.  If you plot the DI of Derby winners by year since 1940 and subject the data to linear regression you will obtain a trend line indicating an average increase of close to 0.04 DI units per year over the last 65 years.  This trend is depicted graphically in Chart 1.

Chart 1.  The Evolution of Pedigree Type in the Kentucky Derby

Extending the trend line 20 years into the future strongly suggests that in the absence of a dramatic shift in breeding patterns we can expect half of all Derby winners by 2030 to have a DI over the previous historical guideline figure of 4.00.  This alone is compelling evidence that the pedigree type of Derby winners is constantly evolving toward greater speed, a development that is troubling to traditionalists who believe excessive speed in pedigrees is a detriment to the Thoroughbred and undermines the meaning of classicity.

But even if we accept that the pedigree type of Derby winners is changing, we are still confronted by the question of whether the changes we see have made pedigree less important to Derby performance.  One way of addressing the question is to attempt a correlation of the quality of individual Kentucky Derbies with the pedigree type of the winners.  How can we do this?

One measure of race quality is through the use of speed figures that tell us how fast a race was actually run.  Speed figures are preferable to the raw final time of a race because time can be dramatically affected by track speed on a given day.  Perhaps the longest established, widely known and most reliable speed figures available to the general public are those published by Andrew Beyer, long-time racing columnist for the Washington Post.  Beyer, an internationally respected turf writer and handicapper, has been publishing his figures for quite some time, providing us with a good historical record.

The complementary historical pedigree type of Derby winners is available using Dosage figures.

The following table displays the DI of all Derby winners since 1991 along with the Beyer Speed Figure (BSF) for each year's Derby.  We use 1991 as the starting point because that year was the first in which the historical DI 4.00 guideline was exceeded and it represents a dramatic shift in the pedigree type of Derby winners.  The DI used in the table is the DI of the winner at the time of his Derby win and does not reflect any subsequent changes resulting from the addition of new chefs-de-race. 

Chart 2 plots the DI against the BSF.  The red line is the trend line derived by linear regression.

Chart 2. The Relationship Between Derby Winner DIs and Their BSFs

Although we see the expected typical scatter in the data, the trend line is quite revealing.  The slant of the line from upper left to lower right indicates that the higher BSFs are associated with lower DIs.  Said another way, Derby winners with more stamina-oriented pedigrees tend to run faster and earn better speed figures.  Of course there are exceptions, such as War Emblem with DI 3.40, BSF 114 and Sea Hero with DI 1.12, BSF 105.  However, the purpose of a statistical analysis is not to show that every member of a population with a common characteristic fits the population profile,  Rather, the purpose is to demonstrate whether separate populations, each sharing unique characteristics, are the same or different.  In this case we are trying to determine if Derby winners with higher DIs (i.e., relatively more inherited speed in their pedigree) have exhibited a different level of performance in the Derby than winners with lower DIs (i.e., relatively more inherited stamina).  Note in this study that there are two clusters in the data, one group (or population) with DI 3.00 and above and another group (or population) with DI below 3.00.  A statistical t-test showing whether the two populations are different from one another by more than mere chance yields the following result when we compare the BSFs of the two groups.

Perhaps most important to this analysis is the P-value of 0.02, suggesting that even though the sample sizes are relatively small, the difference observed between the two groups is statistically significant.  It tells us that there is only about a 2% chance that the difference we see in the average BSFs for the two groups is the result of chance.  For reference, a P-value of 0.05 or less indicates statistical significance.

We can extend the analysis to include the other American classics, the Preakness and the Belmont, to see if the observations made for the Kentucky Derby apply more broadly.  The following chart shows the DI-BSF relationship for all three American classics over the last decade and a half.

Chart 3. The Relationship Between Classic Winner DIs and Their BSFs

In fact, the correlation between DI and BSF is essentially the same as observed for the Derby alone.  

From this brief analysis we may interpret the result to mean that pedigree actually does still make a difference in the Kentucky Derby as well as the Preakness and Belmont, not necessarily in terms of which horse will win, but rather in terms of the quality of the classic performance.  We have seen several classic winners with DIs over 4.00 in the last few years, but they belong to a group of speedier-bred horses that have not performed as well as their stouter-bred counterparts.  Apparently not all classic races are created equal, and at least in recent years that inequality correlates especially well with the pedigree type of the winners.  Stamina, required to stay a classic distance in classic horse time, is still a great asset.  When a horse not traditionally bred to stay a classic distance wins a classic race, there is a reasonable probability that the race will fall short of historical classic standards.