The Receipt Is Older Than the Result: Jepchirchir, EPO, and the Column Sydney Never Checked

30 August 2026, Sydney. The first women's title of the seventh World Marathon Major. Peres Jepchirchir crosses the line in 2:18:31, and Kenya's own Irene Chepet Cheptai finishes three minutes and four seconds behind her. The broadcast had its story ready: Olympic champion, world champion, first winner of a new major, and all of it inside three years.

That evening one column in my spreadsheet stayed empty: when was the sample taken. The answer came later from the Athletics Integrity Unit — an out-of-competition sample collected in Kenya, before the race, tested positive for EPO. Which means that at the moment the gun went off in Sydney, two results were running in parallel. One on the road, one in a laboratory. The market, the broadcaster and the sponsor paid for only the first, because the receipt on the second had not yet been printed.

This report audits that sequence. A positive test is not new. What is new is that the receipt is older than the result.

Context: a résumé written in dates

Jepchirchir, 33, sits in my file as line items, not as sentiment. A women's half-marathon world record in 2026; the same record broken three more times in 2026, a year in which the global calendar was collapsing. 7 August 2026, Sapporo: Olympic gold in 2:27:20, ahead of Brigid Kosgei and Molly Seidel, in heat where three women finished inside 2:28. New York 2026, Boston 2026, London 2026 — and in London, by my note, 2:16:16, the women-only mark, which I always print with the qualifier attached. September 2026, Tokyo again: the world title. Then Sydney.

The résumé is what decides the size of this case. A positive sample from the fortieth-ranked marathoner is a line item. A positive sample from a reigning Olympic and world champion reprices the entire circuit — appearance fees, broadcast rights, a major's branding, and the whole Kenyan distance-running economy that runs on that one reputation.

My method is old and I repeat it. Athens 2026, the Olympic men's long jump market, handed to me at 38, the newest man in a six-person syndicate on the first floor of an office on Smithdown Road, Liverpool. I started from IAAF season lists and wind readings, and flagged Anju Bobby George's six finals beyond 6.60m that summer as a top-eight outlier that the London books were pricing on reputation. She finished fifth with a 6.83m national record and the syndicate cleared £11,400. The Athens outlier was not noise; it was an invoice, and someone had to pay it. From that week I have refused to publish a number without its conditions attached — wind, altitude, sample size, competition level. Every preview opens with the measurement environment before it opens with the athlete.

In this case the measurement environment is the Athletics Integrity Unit, operating since 2026 as a body independent of World Athletics: it selects the tests, issues the notice, imposes the provisional suspension. A provisional suspension is not a ban, and the difference is not lawyerly pedantry — it is the spine of the process. The athlete may request analysis of the B sample; if the B sample confirms, the case goes to a disciplinary tribunal; the tribunal fixes both the sanction and the date from which results are disqualified.

EPO sits in category S2 of the WADA Prohibited List — peptide hormones and growth factors — which means it is prohibited at all times and does not wait for a season or a competition. As a non-specified substance, the standard sanction is four years unless the athlete establishes contamination or fault.

And the physiology matters, because the whole logic of the case sits inside it. Erythropoietin is a natural substance produced in the kidneys that stimulates the creation of new red blood cells. Injected, it raises red-cell volume and the oxygen-carrying capacity of the blood, which is the most expensive commodity over 42.195 kilometres. It is detectable in urine or blood only for a short window. That short window is not a footnote — it is the architecture. Where the detection door is that narrow, the outcome depends entirely on which day, in which city, in which hotel lobby the collector arrives.

Core: four ledgers, and only one of them was argued about

Ledger one — the race arithmetic that kills the easy conclusion

2:18:31 is 8,311 seconds. Across 42.195 kilometres that is 3:17 per kilometre. The winning margin, 3:04, is 184 seconds, or 4.36 seconds per kilometre. Three numbers, one very clear picture — and it is not the picture social media painted.

First, 2:18:31 is not a fast time by 2026 standards. Her own London 2026 run of 2:16:16 is more than two minutes quicker. Sydney's course, an August in the southern hemisphere, running alone — all of it explains the clock. So the line "she ran too fast" does not stand up in the numbers. That matters, because the laziest shape of an accusation is always the same: the time looked suspicious, therefore the sample looked suspicious. There is no arithmetic relationship between the two.

Second, a three-minute margin is unusual in a major, where wins are normally settled in seconds. But a margin is an artefact of race shape, not proof of capacity. Sydney's field was built around one name; the second woman was also Kenyan and had her own calendar. When a field contains exactly one athlete of pacing-group quality, the gap widens. Treating the margin as evidence is the same error as treating a fourth place as a moral victory — writing emotion into the ledger.

Third, and most important: what the margin proves is not about performance but about purchasing. The organiser bought a name and got a solo show — three minutes clear, no drama, no sprint finish, ideal for television. Which is why the damage here is commercial as much as competitive.

Ledger two — eligibility, not performance

I trust the receipt more than the highlight reel. An out-of-competition sample collected before the race does not reach backwards and explain the race. It suspends the right to have been in the race. Merging the two is the mistake we kept making during the 2026 relay cases, when one scandal was allowed to stand in for a decade.

Under the World Athletics Anti-Doping Rules, an established violation normally disqualifies results from the date of sample collection, and a tribunal may reach further back. By that rule, the Sydney title, its prize money and the majors' series points are all directly exposed.

By the same rule, the Tokyo 2026 Olympic gold, the September 2026 world title, London 2026 and Boston 2026 are not automatically exposed. No finding covers them unless a panel pulls the date backwards. And the four half-marathon world records? No rule touches them. Only the public's column does.

That asymmetry is the one I keep separately. The market prices taint immediately; the rules price it slowly; and exoneration never trends. An accusation travels the world in twenty-four hours and its answer arrives eighteen months later, by which time the camera is at another race.

Ledger three — the Jeptoo file, which tells you what happens next

I prefer a comparable before a forecast, and the closest comparable is Rita Jeptoo. An out-of-competition sample collected in Kenya in September 2026 came back positive for EPO. The notice became public after she had won Boston in April 2026. A two-year ban was imposed in January 2026; she appealed; in 2026 the Court of Arbitration for Sport raised it to four years, and the Boston title went with it. The lesson is not the legal fee. The lesson is that contesting cost two extra years.

The second comparable is Jemima Sumgong: Rio 2026 marathon champion, an eight-year sanction in 2026 on an EPO case. On my record her Rio gold was not stripped, because the violation was dated after that race. That detail is the whole argument of this piece: the date decides which medal travels and which one stays. Which is why I print every asterisk with its date attached, and refuse to treat an undated asterisk as a verdict.

Those two give a frame that applies directly here: the sample date, the B-sample confirmation, the tribunal panel, the length of sanction, the date of disqualification, the return of prize money. Those are the five columns that will be filled over the next twelve months. Anyone forming an opinion before them is forming an opinion without the arithmetic.

Ledger four — payroll, appearance fees, and the risk an organiser bought

State the structure before the comparison, because comparing without it is the habit that comes easiest from a London desk. Indian athletics has no league; its money arrives through sports-quota employment — Railways, the armed forces, state police, PSUs — topped up by TOPS and Khelo India. That architecture puts a salary floor under the athlete, which is why disciplinary pressure in India is centrally applied. Britain's architecture is different again: UK Sport's World Class Programme issues centralised grants and can withdraw them.

Kenya has neither. Its distance-running economy runs on management-agency advances, shoe-company retainers, appearance fees and prize money. The athlete is a sole trader, not an employee. A provisional suspension there is not a salary stop but an income stop, with no institutional cushion underneath and no union beside it. That is why the first phone call in these cases goes to the agent, not the federation.

Now the transfer-window part, because this is at root a transaction. The majors are a market in which seven races bid against each other for names. Sydney became the seventh major in 2026; in 2026 it had to prove the promotion, and it bought the reigning Olympic and world champion. Appearance fees are the real money in this market, not prize money — at the top end, a runner earns more for starting than for winning.

Which raises the question that should be put to the organiser, not the athlete: when you buy a name, who carries the residual risk on that name's sample? Is there a doping warranty in the appearance contract? Is the fee escrowed or advanced? What does the restitution clause say if the result is annulled? And most basic of all — is the pre-race out-of-competition screen funded by the race or by the AIU? I have kept that question in my column since 2026, when a newsroom of 22 writers all meant something different by expected goals and I had to issue a 40-page metric dictionary and make adherence a condition of publication. A metric dictionary is a promise you can audit. So is an appearance contract.

Ledger five — latency is the metric nobody measures

EPO's detection window is short, which makes the highest-yield testing location in the sport exactly where this sample was taken: in Kenya, before competition. The system worked. The test was selected, the laboratory found it, the notice went out. What failed was not testing but timing. The gun went off while the screen was still in a lab, and a new major declared the result final within minutes.

My proposal is not a longer ban; the statute already fixes that. My proposal is that latency be measured like a metric — days from sample to notice, published. A circuit that has expanded to seven races has not expanded its verification layer at the same rate. A seven-race market that calls results final on the day is underwriting its own product quality.

Here I should date one assumption, because an undated assumption is not a promise but a habit. My assumption, held since 2026 when the AIU's intelligence-led model took shape, is that positives like this are usually the last page of a long file rather than a sudden accident. The way to falsify it is simple: publish test counts and outcomes by country and event. On the day that data lands, I will re-date the premise.

When the sample is small, the process must be loud. Both senses of the word apply here.

Contrarian: the collective verdict is the cheapest commodity on the shelf

The line that travelled fastest the next morning was: another Kenyan, the marathon is rotten. I will not print it, and the reason is arithmetic. A base rate is not a finding. Kenya's distance-running population is large enough that any positive drawn from it can be projected onto the whole population — the statistical move that always works and never proves anything. Each athlete gets a separate case, a separate sample, a separate date, a separate panel.

And I have to apply the same discipline to the defence. A short detection window means a positive is strong evidence and a negative is weak evidence. The direct implication is that dozens of clean tests are not a character reference — they are a record of the days when a collector was present and the window was open. The argument I refuse the prosecution, I refuse the defence as well.

The third contradiction is the least comfortable. In this case the greatest damage has been done to the part of the résumé no rule can touch. The Sydney title sits in front of a panel; London 2026, Boston, New York, the two Tokyo titles and four half-marathon world records do not. Yet the market has discounted all of them, and that discount will not lift, whatever the B sample says. I keep a column for what the market refuses to price — and here it is the mirror: the market prices the taint before the rules do.

And the fourth, the one that irritates me most. One clean number — a three-minute margin — is enough to build a moral story on. Four decades on this beat have taught me how a single figure swallows an argument. So put a depth column beside every headline mark: what the second-placed time was that day, how many women in the field were inside 2:25, how many of her eight major starts ended in the top three. Fill those columns and the story turns out to be about the thinness of the field, not the width of the margin.

Takeaway: five dates to watch

The B-sample analysis is the first gate, usually weeks rather than months, though EPO queues can run long. Second, whether the athlete contests the provisional suspension — the Jeptoo file shows that the price of contesting is paid in years of sanction. Third, the tribunal panel and its reasoning. Fourth, and most expensive, the disqualification date: 30 August 2026, or something a panel pulls backwards. Fifth, the Sydney 2027 contract — the next time a new major buys a name, will it buy anything about the sample as well?

Someone pays for the gap between the gun and the laboratory. The only open question is whether the invoice lands on the athlete, on the race, or on a circuit that has grown to seven races and still does not publish its verification arithmetic. The next major is signing its 2027 headliner now. The answer is needed before the signature.