Breeding — A Starting Expectation

Pedigree is most useful where direct evidence is thin.

Theory: relatives help frame the question

A sire is the horse's father; the dam is its mother. The damsire is the dam's sire. Their records and those of related runners can suggest likely strengths in distance, surface and development.

These are tendencies, not instructions the horse must obey. Inheritance is only part of the story. Training, health, temperament and opportunity affect the performances we observe.

Pedigree is especially useful for a debutant, a lightly raced horse or a runner attempting an unfamiliar task. When a horse has already supplied repeated relevant performances, those observations usually deserve more weight than a general expectation about its relatives.

Start with today's missing information

Suppose a horse has only raced on turf and now tries dirt. Its turf rating tells us something about demonstrated ability, but not exactly how that ability transfers.

A sire whose offspring have performed well on dirt can support the possibility of suitability. Then inspect the horse's own movement, experience, likely pace position and preparation. Pedigree does not establish kickback tolerance or current fitness.

For a longer-distance attempt, ask whether the family provides a plausible starting expectation of stamina. Then check how the horse has used its effort in shorter races. A strong finish after an easy early pace is different evidence from sustaining a demanding race.

Read the denominator

“Two hundred starts on dirt” could come from many offspring or repeated starts by a small group. The number of distinct horses matters, as do age, class and where those starts occurred.

A sire's turf and dirt averages may compare different populations. Better offspring might be sent to one country, ordinary ones to another. Differences in opportunities and selection can look like a surface preference.

Inspect both the size of the effect and the breadth of the sample. A small difference in average rating is not automatically meaningful simply because a table reports a decimal. We have measured all three ways a sire table misleads: the same stallion reading positive in one country and negative in another; apparent sire edges that were a handful of elite yards in disguise, collapsing once those trainers were set aside; and sire signals that faded to nothing over eight seasons as the market caught up. A sire opinion is local and perishable, so check the date on it.

The sire effect you want to understand and the betting effect you want to exploit are also different. A familiar sire's reputation may already be strongly reflected in the odds.

Do not count the label and its ingredients twice

A surface badge or pedigree score may summarise the same starts shown in a statistical panel. The badge and the panel are one source expressed in two ways.

If a model already uses sire information, a manual pedigree adjustment should address something more specific than “this sire likes dirt.” Perhaps the displayed grouping is too broad, or the individual horse has relevant evidence that the model handles poorly. If you cannot identify the added information, keep the original estimate.

The same applies to overlapping relatives. A sire, a sire line and a related pedigree description are not necessarily independent confirmations.

Surface groups are convenient, not biological identities

Some data systems combine categories to gain sample size. That can be useful for modelling while concealing differences important to an individual horse.

Check what a turf, dirt or all-weather category actually contains. A success on one synthetic surface does not establish equal suitability on all synthetic tracks, and an apparent dirt preference may reflect the particular racing population in the sample.

An experienced horse's own record under comparable conditions should challenge a broad pedigree label. The label is there to help when evidence is missing, not overrule the evidence that arrives.

Pro tip

There are no mudlarks, only mud families

On going specifically, the family told us more than the individual. Once a horse's career trajectory was removed, its own going preference nearly vanished: a point or so at most, against eleven-plus points of ordinary race-to-race noise. The horse that loves soft is very largely a story built on two wet wins. Sire-level wet aptitude, by contrast, was real, because twenty or more progeny can show a trait that no single horse's dozen starts can, and in Europe it looked underpriced. So when the rain comes, spend less time on what the horse has done on soft and more on what the sire's stock does on it. This is a going-specific exception to the general rule above that own-horse evidence deserves the greater weight. (MWP research; the horse-level result high confidence, the sire-level result medium.)

Development without a fixed timetable

A family associated with later development may justify patience with an inexperienced runner. It does not prove that a particular horse will improve at four, or that an early winner has finished progressing at three.

Use pedigree to widen or shift your expectations cautiously. Then update them as the horse supplies its own races. The value of a prior expectation is that it can be revised.

Put uncertainty into the price

A first-time surface switch can create a wider range of possible performances. The horse may transfer its ability, improve or struggle. A big price is not attractive solely because the uncertainty is large.

Conversely, an unfamiliar task does not make every short price wrong. The available evidence may be strong enough to support a substantial chance. The question is whether the price adequately reflects your range of outcomes.

You should be able to explain which part of the estimate comes from the horse and which part comes from relatives. If you find yourself replacing a weak personal record with an elaborate family story, pause.

Practical: own evidence versus family evidence

Quiz: pedigree in proportion

Up next: the model combines many inputs. Your job is to understand its output before changing it.