What Inbreeding Does to a Thoroughbred's Racing Record

Every Thoroughbred racing today, in the Kentucky Derby field or at a Tuesday card anywhere in the world, traces back to the same three stallions: the Byerley Turk, the Darley Arabian and the Godolphin Arabian, imported to England between the 1680s and 1729. The breed's stud book, the General Stud Book, closed to outside blood in 1791 and has stayed closed since. Some degree of inbreeding, in the sense of two horses sharing an ancestor somewhere in their pedigree, is close to unavoidable after 235 years of that. The real questions are how much it matters and to what degree.

The largest study to measure it, a 2018 genetic analysis of over 135,000 Australian-bred runners, found that it does matter: horses with a higher inbreeding coefficient earned less and started fewer races, on average, than less inbred horses from the same era. It also found the effect was not uniform. Some of the breed's most-repeated ancestors were linked to worse performance in their descendants; at least one was linked to better.

A horse is inbred to a given ancestor when that ancestor appears more than once in its pedigree, on the sire's side and the dam's side both. Geneticists measure how much of this has happened with an inbreeding coefficient: roughly, the chance that two copies of the same gene, one inherited from each parent, both trace back to that one shared ancestor rather than to two different ones. A higher coefficient means more of the genome is doubled up this way, for better or worse, rather than drawn from a wider set of ancestors.

Breeders often call a distant repeat (the same ancestor four or five generations back on both sides) "linebreeding" and a close one (two or three generations back) "inbreeding," but it is the same mechanism at different intensities: the closer the repeat, the larger the share of the genome it accounts for.

In 2018, a team led by Evelyn Todd at the University of Sydney published the most thorough look yet at inbreeding and racing performance, in the journal Scientific Reports. They traced the pedigrees of 135,572 Australian Thoroughbreds that raced between 2000 and 2011 back through a genealogy of 257,249 individuals, and tested whether a horse's inbreeding coefficient predicted its earnings, its number of starts and three other performance measures.

It did, in every one of the five measures, and the relationship ran the same direction each time: more inbred horses performed worse, on average. The effect was not the same for every ancestor, though. Just 10 individuals accounted for more than 80 percent of the matching gene copies (what the paper calls identity-by-descent alleles) carried by the modern population, led by the 18th-century sire Herod at close to 20 percent on his own, then Eclipse and St Simon. Horses inbred to Herod actually performed somewhat better; horses inbred to Eclipse, Stockwell or Touchstone performed worse. That split is the paper's central point: inbreeding itself is not uniformly good or bad, because it is really a question of which specific, centuries-old genetic variants are getting doubled up, and some of those variants are more harmful than others.

No stallion is more often named in this conversation than Northern Dancer, the Canadian-bred champion of the early 1960s who became one of the most influential sires in the breed's history. He is not one of the three foundation sires, but by the time his career as a stallion ended, his descendants had spread so widely through the commercial breeding industry that avoiding him completely in a modern mating, especially an American one, is difficult by design rather than by accident: breeders have spent 60 years deliberately breeding to his sons, grandsons and great-grandsons because they produced good racehorses, and each of those matings added another path back to the same horse.

That is the practical version of the 2018 study's finding. A closed stud book of a few hundred thousand living horses, bred for a century and a half toward the same handful of outstanding sires, was never going to stay free of repeated ancestry. The question for a breeder, and the one the paper was built to answer, is which repeats are worth the risk and which are not.

For a bettor, not much on any given race day. The 2018 study's effect is a population-level average across thousands of horses and eleven years of form, not a reliable way to size up one horse in one field; a program page does not print an inbreeding coefficient, and calculating one by hand from a catalog pedigree takes real pedigree software. It is a better fact to know for reading the yearling sales or a stud-fee announcement than for reading a past-performance line.

What it is worth taking from this is the plainer point: a pedigree stacked with the same famous names is not, by itself, a mark of quality. It is a mark of fashion, and fashion and soundness are not the same thing.