Live data — cohort re-joined to official race results nightly | Claims last calibrated 22 July 2026

Know which yearlings to buy Buy the yearlings built to win.

We measure how a yearling walks. The ones we rate highest race longer — and return 28% more prize money per dollar — than horses bought for the same money.

The eight numbers

The Eight Furlong Edge.

One standard behind every number: mature horses with complete race records, top-rated compared against bottom-rated bought for the same money. Verified against complete career records as at 22 July 2026. Careers are still running, so the cohort refreshes nightly and these numbers are re-calibrated on a fixed cadence — never left to drift silently.

01 · Marquee
+28%

More prize money per dollar

Top-rated yearlings returned 28% more prize money per dollar than bottom-rated bought for the same money. It comes from starts, not speed: +15% more starts per dollar, while earnings per run are statistically flat.

n = 5,233 · yearling-sale, verified
02 · Within-sire
+28%

Same sire. More return.

Choosing between two foals of the same sire? Across 130 sires, our top-rated progeny returned 28% more per dollar — pedigree and price held constant. The one number that beats both "you just bought cheap" and "you just bought the right sire."

130 sires · same sire, same budget
03 · Durability
1.25×

Still racing at 30 starts

18.8% vs 15.1%. Top-rated yearlings are ~25% more likely to still be racing deep into their careers — when the real money is paid. Soundness is the edge, and it's our most robust number.

n = 5,233 mature careers
04 · Deep career
1.98×

Nearly twice as likely at 50+ starts

The edge compounds the longer a career runs. Top-rated yearlings are 1.24× more likely to reach 30 starts, 1.67× at 40, and 1.98× at 50+. Soundness is what we measure, and soundness is what shows up when the career goes long.

n = 5,233 · 95% CI 1.29–3.22×
05 · Both sexes
+25% / +21%

Fillies / Colts

The value edge holds on both sides of the catalogue — +25% for fillies, +21% for colts. Not a fluke of one population.

same standard, split by sex
06 · Stability
+6–30%

Across three complete vintages

The prize-money-per-dollar edge is positive across every yearling vintage with mature careers (2020–22). A repeatable signal, not a single-year fluke.

2020–2022 · same standard
07 · Rarity
1 in 16

Net +6 = top 6.2%

The signal is rare on purpose. Net +8 = top 1.5% / 1 in 66. Few horses clear it — and that's the point.

whole cohort · biomech score only
08 · Same sire, races longer
1.57×

Same sire. Longer career.

Every stallion produces sound ones and fragile ones. Compare their foals side by side and our top-rated picks are 1.57× more likely to still be racing at 40 starts than the bottom-rated from the same sire. Same catalogue page — pedigree and price held constant. Biomech doing the work.

130 sires · 95% CI 1.23–2.00×
What the warning actually is
“It can't warn you off a slow horse. It can warn you off a fragile one — bottom-rated yearlings are markedly less likely to reach 40–50 starts.”
That's a soundness warning, not a talent warning — and soundness is what decides whether a yearling ever repays its price. Top-rated: 15% more starts per dollar, 1.67× more likely to reach 40 starts, 1.98× at 50+.
What it looks like in practice

$6,000 in. $493,875 out.

PERSIAN SPIRIT — 2022 Premier Yearling Sale, Lot 665. We rated him net +6 on the walk — top 6% of the cohort — before he ever set foot on a racetrack. He has since won 6 of 22 starts, reached stakes company, and returned 82× his purchase price. Below is the actual single-horse report the model produces — the same one you'd get for a lot you're considering.

Read the full Persian Spirit report →

The data

25,010 yearlings. Measured.

The data asset behind every claim on this page — measured, then plotted.

Reach
25,010

Yearlings measured

Every yearling we've put through the model — Inglis, Magic Millions and NZB, 2019 to 2026. The data asset behind every score.

operational cohort
Validation
5,233

Racehorses, full careers

The claims aren't modelled — they're measured against 5,233 mature Australian racehorses with complete career records, bought as yearlings, identities verified. Real prices in, real prize money out.

AU validation cohort

Every horse in the cohort, plotted on two of the biomech metrics that drive the score.

Distribution · Leg-to-body ratio
Spread · Head height vs Leg-to-body ratio
Colts & geldings Fillies & mares
Where we draw the line
“We publish one signal, because one signal survived the testing. Soundness holds under every test we run — so that is what we sell you.”
Top-rated horses do earn ~11% more per start, but that gap isn't statistically significant — so it stays off this page. The discipline is the product: if a number can't survive being tested on horses the model never saw, we don't print it.
Talk to us

Bring this data analysis to your next sale.

Briefings for bloodstock managers, syndicators and trainers. Eight Furlong is private and bespoke — we don't publish picks.