The Number I Nearly Published
Ninety-nine cents.
That's the figure I had ready to lead this piece: the median foal returns 99 cents per dollar of service fee, across every Australian stallion where both numbers can be measured. Half of all rankable sires return the fee at the median foal; half don't. A clean parity story. It was drafted, charted, and a day from publication.
It was also wrong, and the error is worth walking through, because the corrected number tells a different story.
The setup
I joined the advertised 2026 service fees of 288 stallions to the two-season race database behind the earlier pieces in this series: every runner, every dollar actually paid, seasons 2024-25 and 2025-26. For each of the 148 sires with enough runners to rank, I computed what the median foal earned and divided it by the fee. Median across all of them: 0.99. Half above, half below.
The problem is in the units. A service fee is paid once. The earnings window is two seasons. But a racing career is not two seasons long, and comparing a one-off cost to a two-year slice of a longer income stream understates the return on every sire in the table.
How long is a career?
The data can answer this, because it holds every runner's foaling date. Two estimates, from two directions. The age distribution of runners, treated as a survival curve, gives 3.99 seasons for a horse that ever starts. The demographic arithmetic (28,093 runner-seasons per season, against a foal crop of 11,691 from the Racing Australia Fact Book) gives 3.0 to 4.4. I've used 3.66, the midpoint, and every adjusted number below scales linearly if you prefer a different point in that range.
There's an offsetting correction. The fee is paid on every foal, but not every foal races. Racing Australia's own cohort tracking found 51% of the 2014 crop had raced before their 4yo season, and a Victorian cohort study (Australian Veterinary Journal, 2022) found 64% of foals raced at least once; together they support a realistic range of 55% to 72% ever starting. I've carried it as a range, midpoint 65%, because the database genuinely cannot measure horses it never sees.
Career length multiplies the return by 3.66. The starter rate multiplies it by 0.65. The first dominates.
The corrected numbers
Career-adjusted, per foal actually bred: 76% of rankable sires return the fee at the median foal. The median rankable sire returns $2.26 per fee dollar. The parity story was largely an artefact of the truncated window. Stallion fees, at the middle of the market, are cheaper relative to median racetrack earnings than the naive analysis suggested. Meaningfully cheaper.
What survives untouched is the top of the fee market. Zoustar at $275,000 moves from 9 cents per fee dollar to 21 cents per foal, career-adjusted, with 10.8% of foals clearing the fee. Extreme Choice moves to 14 cents. The correction triples their multiples and they remain deeply negative on racetrack return alone. That's not a scandal, and the reason bears repeating: six-figure fees buy a chance at the star outcome, and the return arrives in the yearling ring, not the prizemoney account. This analysis measures what a breed-to-race operation collects, and at the top of the market, that operation collects a fraction of the fee at the median.
The value end amplifies. Akeed Mofeed ($3,300) returns 8.2 times the fee per foal, career-adjusted, with more than half his foals clearing it. Divine Prophet ($5,000) returns 8.0 times. Two large-sample names join them at the top of the career table: Your Song at 11.2 times across 137 runner-seasons, and Headwater at 10.4 across 401. These are not fashionable sires, which is largely why the arithmetic works.
The concentration warning from the last piece still applies: a sire can sit on the value list because one outlier inflated his numbers, and any multiple read without its concentration figure is half a number. The dashboard carries both columns.
A curve worth knowing
One byproduct of the age analysis deserves its own line. The 2yo season carries 6.2% of all prizemoney. The peak is the 4yo season, at 25.9%, where the median runner earns $17,235. The sales ring prices horses at age one and races them at two. The money arrives at four. Anyone modelling cash flow on a racing investment should have that curve taped to the wall.
Explore it yourself
The full dataset (188 sires, raw and career-adjusted multiples, break-even rates, concentration, age and distance profiles) is now live as an interactive dashboard at shaunalytics.com/sire-roi. The naive columns sit alongside the adjusted ones, because showing the correction is the point. Sort it, filter it, check my working.
Notes
Fees are advertised 2026 service fees (Breednet); earnings are the 2024-25 and 2025-26 seasons, prizemoney as paid. Career adjustment: per-runner-per-season median earnings, multiplied by an estimated 3.66-season career (two convergent estimates documented above; foaling dates complete for 100% of runners), multiplied by a 55-72% starter rate (Racing Australia cohort figures and a 2022 Australian Veterinary Journal Victorian cohort study; carried as a range, midpoint shown). The adjusted figures are projections built on a stable-population assumption, not observed careers; the assumption and its biases are documented in the methodology. Prizemoney only: no sale proceeds on the return side, no rearing costs (typically $30,000-50,000 a year before a horse races) on the cost side. Multiples therefore describe racetrack economics, not total breeding economics. That gap is largest at the top of the fee market: well-bred horses, especially by premium sires, carry residual value beyond racing (stud careers for the best colts, broodmare value for fillies) and none of that appears on the return side here.