Dosage dates from the early 1900s and family numbers from the 1890s. Both were drawn up by hand, and neither was built by checking what yearlings went on to do. The HorseBytes System starts from that check: it looks at what earlier yearlings and foals in the HorseBytes database achieved, and keeps only the parts of a pedigree that told the good ones from the rest.
In the test, yearlings and foals in the top fifth of the index won black type 6.3% of the time. Those in the bottom fifth: 0.7%.
On a scale where 0.5 is a coin toss, the index scored 0.66 at picking later black-type winners. Dosage points scored 0.51 on the same horses.
The HorseBytes Index. A lot with an index of 80 has a stronger pedigree than 80% of the young horses in the database, with the relatives that lift or lower it shown alongside.
Every pedigree theory judges a horse by its relatives. The HorseBytes System does it the way animal breeders do: it measures each relative on the same scale, checks how much each one has mattered in the past, and adds them up with those weights. Each part is compared with the average young horse in the database, so a sire with twenty black-type winners and a 2nd dam with three are put on the same footing.
The number beside each part is its weight: how far one step above average on that part moves the index. Red lowers it.
A catalogue page lists a dam's best foals and leaves the moderate ones out, so averaging the horses it names flatters every family. Black-type winners are always named, so that is what the HorseBytes System counts.
A sire's black-type winners count only if they were born at least two years before the lot: what a buyer could have known on sale day, not what the stallion did later.
Every part had to earn its place by picking out later black-type winners. Ideas that didn't — the dam's number of winners, the class of the page, a high dosage count — were left out.
We went back through 25 sale catalogues from 2002 to 2026 and found 2,242 foals, yearlings and two-year-olds whose dams turn up again in a catalogue at least four years later — long enough for any black type to be printed under her. 77 of them (3.4%) went on to win black type. Each horse was given the index it would have had on the day it was sold.
Average 3.4%. Each step down the index, fewer went on to win black type.
One of the strongest signs in the data — and one that dosage and family numbers ignore.
| Measure, on the same horses | Score (0.5 = guesswork) | Top fifth won black type | Bottom fifth |
|---|---|---|---|
| HorseBytes Index | 0.66 | 6.3% | 0.7% |
| Dosage points | 0.51 | 4.2% | 3.3% |
| Class of the catalogue page | 0.51 | 2.2% | 2.7% |
| Dam's winners from runners | 0.44 | 2.9% | 1.8% |
The score is the chance that, taking one later black-type winner and one non-winner at random, the measure rates the winner higher. 0.5 is a coin toss and 1.0 would be perfect. Pedigree alone can never come near 1.0 — the horse itself, its trainer and luck matter more — so a clear lift above 0.5 is what counts. The weights were set on these catalogues, so the HorseBytes app re-runs the test on every catalogue added; the more there are, the sterner the check.
| Dosage and family numbers | The HorseBytes System | |
|---|---|---|
| Where it comes from | Lists of influential stallions and founding mares drawn up by hand, decades ago | The HorseBytes database, re-counted every time a sale is added |
| Tested? | Rarely, and then on the horses it was built from | Yes — on horses whose later careers are known |
| Up to date | A stallion counts only once he joins the list | A young stallion counts as soon as his first black-type winners appear |
| The close family | Dosage looks four generations back on the sire's side; family numbers look 300 years back on the dam's | Weighs the sire, dam, 2nd and 3rd dams and the dam's sire by how much each has mattered |
| The dam's age | Ignored | Included — one of the strongest signs there is |
| What it tells you | A likely trip and type | How likely the pedigree is to produce a black-type horse, and which relatives lift or lower it |
Every weight comes from counting what happened to real yearlings, so anyone can check the result — and the check is repeated as the database grows.
Catalogue pages show off a family's best foals. Counting only black-type winners, which every page names, keeps a well-written page from flattering a moderate family.
A stallion is judged on what he had done when the lot was born, so a yearling is never credited with winners that came along afterwards.
Every index comes with the relatives that lift it and the ones that hold it back, so you can see why a lot rates as it does.
They answer different questions. Dosage and the family's winning distances are still the best guide to trip and type, and HorseBytes family lines show where a mare comes from and how strong her family is. The HorseBytes Index answers the question a buyer asks first: how good is this pedigree likely to be? HorseBytes reports show all three side by side.
Conformation, size, walk and vet reports aren't on a catalogue page, and they matter as much as the pedigree. A catalogue has no prices, so the index can't say what a lot will make or whether it is good value. And it is a guide to chances, not a forecast: most lots in the top fifth will still not win black type — they are simply far more likely to than the rest. Nothing on this page is a prediction of any horse's ability or value.
Spotted something we've got wrong, or want a topic explained? We'd like to hear from you.