In brief:
Racing ability is assumed to be genetic in origin, and multifactorial in make-up.
Even though the contributing factors are not individually known, racing ability can be measured and numerically expressed. This allows the contributions from sires, broodmares and broodmare sires to be quantified.
For consecutive generations of offspring, racing ability is subject to regression to the mean. The mean filial regression towards mediocrity directly proportional to the parental deviation from it.
The multifactorial origin of racing ability makes conventional terminology like ‘inbreeding’ less than useful, as it is not known what exactly is being ‘inbred’ for.
It is not known how the individual components of racing ability (to the extent that they may exist) are passed on from parents to offspring, but assumptions can be made.
Karel Miedema – posted January 2022
Racing Ability
Horses compete while carrying weight, at speed, over distance.
This requires the combined effort of different genetic aptitudes.
The effort can be measured by means of ratings, calculated from performance shown in races.
Calculations involve weight carried, as well as racing speed and distance run.
Ratings are expressed in units of weight, and are relative to the rating of an average horse in a given horse population.
In such a population the ratings follow a statistical Normal Distribution.
When taken by sex, males show a greater average ability than females.
The Genetic Lottery
In any racehorse pedigree there are equal numbers of ancestral sires and dams.
A six-generation pedigree, for example, shows 32 each of sires and dams.
The sires, with little exception, are all well above average in racing ability.
The dams can be anything - few have ability of the same level as sires, some are unraced.
Sire and dam contribute equally to the racing ability of their offspring, in random fashion.
The sires will possess a combination of genetic material which facilitates superior ability.
Whether or not they will be able to pass this on to their offspring depends on what exactly they inherited from their own sire and dam. Odds are there will be hidden factors to spoil the broth.
The opposite could be true for dams, able to contribute their hidden factors to good advantage.
So much is random.
As the inheritance of racing ability is multifactorial in nature, the correlation between the measured racing ability of sire and dam is less than perfect.
This should result in ‘regression to the mean’ for racing ability of their offspring in successive generations.
Galileo - Regression At Work
Descendants of champion sire Galileo give insight in the workings of inheritance of racing ability.
To give a benchmark, comparable career-best ratings are shown for broodmares and offspring.
(for explanation of how Ratings are used, see below under ‘References’)
Population totals
| median rating | |
|---|---|
| producing mare | 83 |
| ♂ offspring | 80 |
| ♀ offspring | 74 |
| all offspring | 77 |
Observations:
There is a drop in ratings from mares (83) to offspring (77).
Not all ♀ offspring (74) become broodmares (83) – breeders appear to be selective.
Galileo as sire
| median rating | |
|---|---|
| Galileo’s rating | 132 |
| producing mares mated | 104 |
| ♂ offspring | 96 |
| ♀ offspring | 86 |
| all offspring | 92 |
| Galileo + mares average | 118 |
There is a drop in ratings from mares (104) to offspring (92).
Galileo as broodmare sire
| median rating | |
|---|---|
| covering sires | 122 |
| Galileo mares | 89 |
| ♂ offspring | 86 |
| ♀ offspring | 79 |
| all offspring | 83 |
| sire + mares average | 106 |
Galileo’s producing mares are rated higher than his ♀ offspring (86) – breeders appear to be selective.
There is a drop in ratings from Galileo mares (89) to offspring (83).
The sire x broodmare sire combination myth
Conventional wisdom has it that combinations of sires with certain broodmare sires matter.
For Galileo the combination with Danehill mares is regarded as exceptional, with Frankel the flag-bearer for the cross.
Ratings paint a sobering picture.
Some 15% of Galileo offspring is from Danehill mares.
They have a median rating of 92, which is the same as for Galileo offspring from all mares.
Which suggests that "statistics" involving sire x broodmare sire combinations are flights of fancy.
Regression Is Everywhere
With showing a drop in ratings from mares to offspring, Galileo is not alone.
Regression is everywhere.
Here’s a small sample of sires, to compare.
(G=Galileo; D=Dubawi; DD=Danehill Dancer; S=Shamardal; RA= Royal Applause, P=Pivotal, OD=Oasis Dream, IS=Invincible Spirit)
Sire as sire
| G | D | DD | S | RA | P | OD | IS | |
|---|---|---|---|---|---|---|---|---|
| Sire’s career-best rating | 132 | 128 | 118 | 127 | 121 | 120 | 129 | 120 |
| Rating of mares mated | 104 | 105 | 92 | 99 | 82 | 97 | 100 | 94 |
| ♂ offspring | 96 | 92 | 85 | 86 | 79 | 86 | 85 | 85 |
| ♀ offspring | 86 | 83 | 83 | 81 | 69 | 81 | 79 | 80 |
| all offspring | 92 | 89 | 84 | 84 | 74 | 83 | 82 | 83 |
| sire + dam average | 118 | 117 | 105 | 113 | 102 | 109 | 115 | 107 |
| difference | 26 | 28 | 21 | 29 | 28 | 26 | 33 | 24 |
Sire as broodmare sire
| G | D | DD | S | RA | P | OD | IS | |
|---|---|---|---|---|---|---|---|---|
| Covering sires | 122 | 123 | 124 | 122 | 121 | 123 | 123 | 122 |
| Sire’s mares | 89 | 91 | 90 | 88 | 76 | 87 | 81 | 83 |
| ♂ offspring | 86 | 83 | 83 | 83 | 79 | 83 | 80 | 77 |
| ♀ offspring | 79 | 77 | 80 | 79 | 71 | 78 | 75 | 73 |
| all offspring | 83 | 80 | 81 | 81 | 75 | 80 | 78 | 76 |
| sire + dam average | 106 | 107 | 107 | 105 | 99 | 105 | 102 | 103 |
| difference | 23 | 27 | 26 | 24 | 24 | 25 | 24 | 27 |
In summary:
| Pop. | G | D | DD | S | RA | P | OD | IS | |
|---|---|---|---|---|---|---|---|---|---|
| Sire’s daughters | 74 | 86 | 83 | 83 | 81 | 69 | 81 | 79 | 80 |
| Sire’s producing mares | 83 | 89 | 91 | 90 | 88 | 76 | 87 | 81 | 83 |
| Sire’s mares progeny | 77 | 83 | 80 | 81 | 81 | 75 | 80 | 78 | 76 |
| Difference: dam to foal | -6 | -6 | -11 | -9 | -7 | -1 | -7 | -3 | -7 |
Genius is rare
Who is the best sire?
Sire championships are decided, annually, according to prize money won by progeny.
On a different level, the percentage of stakes winners produced from foals matter.
Both of these are subjective measures, as they depend on the competition from other sires at any given time. Having to compete against weak or strong makes for different outcomes.
Ratings have no such subjectivity, making for a much more objective assessments of ability.
From the results of the most important races, generally designated as ‘Gr1’, a median rating can be calculated.
This way, the median for offspring rated at Gr1 level is estimated at >115 for ♂, >107 for ♀.
Here’s a small sample of sires, to compare their percentage of offspring at Gr1 level.
(G=Galileo; D=Dubawi; DD=Danehill Dancer; S=Shamardal; RA= Royal Applause, P=Pivotal, OD=Oasis Dream, IS=Invincible Spirit)
Sire as sire
| Pop. | G | D | DD | S | RA | P | OD | IS | |
|---|---|---|---|---|---|---|---|---|---|
| ♂ > 115 | 1% | 10% | 8% | 3% | 4% | 1% | 6% | 3% | 4% |
| ♀> 107 | 2% | 10% | 10% | 7% | 6% | 0% | 8% | 4% | 3% |
It shows that in the overall population some 99% doesn’t make it to Genius level.
Outstanding sires like Galileo and Dubawi have about 90% ‘failure’ rate.
Sire as broodmare sire
| Pop. | G | D | DD | S | RA | P | OD | IS | |
|---|---|---|---|---|---|---|---|---|---|
| ♂ > 115 | 1% | 3% | 4% | 4% | 2% | 1% | 2% | 1% | 0% |
| ♀> 107 | 2% | 4% | 2% | 5% | 3% | 0% | 6% | 2% | 1% |
(actual numbers of qualifiers for each sire in this category are small, so percentages should be regarded as no more than tentative indicators).
Beating The Odds
The inheritance of racing ability is multifactorial in nature.
What exactly ‘multifactorial’ consists of cannot be defined at present.
Perhaps several of these factors could behave in the way coat color is inherited (see How Inbred Is Enable), where a recessive chestnut gene can travel unseen for generations, or where a close ancestor’s passed-on colour gene can be eliminated in as little as two generations.
Or mitochondrial DNA could be involved, which is for the most part a female contribution and not passed on by males.
Conventional wisdom of describing horses as ‘inbred n x n to so-and-so’ (like inbred 3x4 Northern Dancer, for instance) has little merit – what is it that the horse would be inbred for, knowing that racing ability is multifactorial and has no specific genetic definition?
Who knows how much of Northern Dancer’s ‘racing ability’ is still present if the stallion is three or four generations removed from the end-product.
So much is random.
Even so, it follows that the closer Northern Dancer would be, the greater the chance of some of his desirable ‘racing ability’ making genes still to be present.
In this context the old adage ‘Breeding the Best to the Best” seems appropriate, even if it is as an exercise in futility.
There is another ‘sport’ where animals are bred for racing ability: racing pigeons.
It is not uncommon for highclass flyers to have relatives close up in their breeding – consider son to mother, father to daughter, nephew to niece.
If racing ability in pigeons is multifactorial, these kind of matings seem likely to increase the probability of combining desired elements in the final outcome.
Similar matings are not easy to achieve with racehorses, mainly through lack of opportunity.
It is possible, however, to achieve ‘close’ matings through individuals whose ancestry shows a high degree of similarity. This can be aptly described as ‘kin-breeding’.
Kin-breeding
Bull Dog & Nogara.
Nijinsky & Mr Prospector.
Just a couple of randomly chosen influences whose ancestral combinations make them ‘kin relatives’. As pairs they can be found in pedigrees of highclass performers.
There are many of such combinations in pedigrees, old and new.
The key is to get to know and record them, then imitate their occurrences. The more the merrier.
If mitochondrial influences count, then it's females that matter – which makes Family Numbers a boon to researchers.
Still, randomness rules in endless possibilities. What appears to work once proves not be repeatable. What seems far-fetched can suddenly surprise.
One day….
Reference material
About the Ratings
A rating is a measure of how a horse’s ability compares to that of the ‘average’ horse in a population.
This means that to determine a true level of ratings, all horses in a population need to be assessed in terms of ability. In other words, all races contested by all horses need to be analysed.
It is imperative that the rating of the ‘average’ horse is kept the same from year to year. This allows for ratings from even decades apart to be reliably compared.
Ratings used in the analysis are from Raceform UK data, for all races run in UK and Ireland, between 1998 and 2021. Performances from other countries have not been used, as Raceform does not rate all races in such countries, which might introduce a measure of bias.
To avoid undue influence of the lowest rated horses (those with little or no racing ability, and thus hard to measure properly) a bottom cut-off rating of 55 has been used for calculations.
The ‘Genius Is Rare’ segment is different - here percentages had to be calculated using ratings of all progeny (no bottom limit, ratings down to lowest).
Understanding Regression to the Mean
(extracted from: Thinking Fast And Slow – Daniel Kahneman; Penguin Books, 2012)
“Regression to the mean was discovered and named late in the nineteenth century by Sir Francis Galton, a half-cousin of Charles Darwin and a renowned polymath.
He published an article in 1886 under the title ‘Regression towards Mediocrity in Hereditary Stature’, which reports measurements in size in successive generations of seeds and in comparisons of height of children to the height of their parents. He found from experiments that offspring did not tend to resemble their parents in size, but to be always more mediocre than they – to be smaller than the parents if the parents were large, to be larger than the parents if the parents were very small. The experiments showed further that the mean filial regression towards mediocrity was directly proportional to the parental deviation from it.
It took Galton several years to work his way from his discovery of filial regression in size to the broader notion that regression inevitably occurs when the correlation between two measures is less than perfect.”
“The correlation coefficient between two measures, which varies between 0 and 1, is a measure of the relative weight of the factors they share. For example, we all share half of our genes with each of our parents, and for traits in which environmental factors have relatively little influence, such as height, the correlation between parent and child is not far from 0.50.”
“It took Francis Galton several years to figure out that correlation and regression are not two concepts – they are different perspectives on the same concept. The general rule is straightforward but has surprising consequences: whenever the correlation between two scores is imperfect, there will be regression to the mean.”
“… extreme groups regress to the mean over time.”
“… regression to the mean has an explanation but does not have a cause.”
Regression to the mean
graphs for sires mentioned in tables in main text body
.png)
.png)
Mitochondrial DNA:
An Important Female Contribution to Thoroughbred Racehorse Performance
(extracted from https://pubmed.ncbi.nlm.nih.gov/16516561/)
Abstract
The mitochondrial DNA (mtDNA) molecule, carrying genes encoding for respiratory chain enzymes, is a primary candidate for demonstrating associations between genotype and athletic performance in mammalian species.
In humans, variation at seven protein encoding mitochondrial loci has been implicated in influencing fitness and performance characteristics. Although thoroughbred horses are selected for racing ability, there have not been any previous reported associations between genotypes and racecourse performance.
The multi-factorial nature of the inheritance of racing ability is an obvious complicating factor.
However, mitochondrial gene variation may represent a measurable component contributing to performance variability.
Previous population studies of thoroughbreds have shown the existence of D-loop variation.
Importantly, we have observed that there is also independent and extensive functional mitochondrial gene variation in the current thoroughbred racehorse population and that significant associations exist between mtDNA haplotype, as defined by functional genes, and aspects of racing performance.
Added:
A haplotype is a set of DNA variations that are usually inherited together.
The alleles making up a haplotype can be located in different places on the chromosome but they are inherited together.
The genotype can refer to a single trait or multiple traits inherited together from a parent.