EMPTY DOSSIERS: ATHLETICS RUNS ON NUMBERS WITHOUT A SOURCE

1. File Number 187

At midnight on 11 January 2026, in Nagoya, I opened a data package of 214 result files sent by a European electronic timing supplier along with a service contract. File 187 had 47 numeric columns. Reaction time: 0.147 seconds. 100m split: 10.31. Wind speed: +1.4 metres per second. Track temperature: 24.8 degrees Celsius. The final column, the one recording the source of the data, was empty.

Not a single line said who the athlete in that file was, which meet they ran, which official signed the technical report, or when the timing system was last calibrated. Yet when I traced it backwards, that 10.31 had already appeared in 11 articles across 6 media outlets, been loaded into two online ranking systems, and been resold by a sports data company to three betting clients in Asia.

A data point with no source travelled more than two thousand kilometres in 61 hours.

I often ask: where did this money come from, and what did it do along the way? For athletics data, the equivalent question is: where was this number born, whose hands did it pass through, and who collected at the final stop?

2. The Route of a Number

A valid athletics result travels through four stages. The first is equipment: the photofinish system or a transponder chip in the shoe, the wind gauge beside the track, the temperature and humidity sensors. The second is people: the backup timekeeper, the wind observer, the official who signs the technical report. The third is the national federation, where a result is checked against the competition permit. The fourth is the international data system, where a result counts only if the meet held the right permit, the timing equipment met the standard, and a wind gauge operated in the correct position.

Any stage can lose the source. A wind gauge set at the wrong angle produces a figure that does not match the direction of the race. A grassroots meet without an international permit can still hold a race, still run electronic timing, still produce a result — but that result counts toward no entry standard at all. An intermediary data supplier has no obligation to check the source: they buy files, process files, sell files.

World Athletics has operated a points ranking system since 2026, in which every result is attached to a tiered meet group. That system only works when the input result carries a meet code, a date, a venue and an athlete code. When those four fields are missing, the software does not stop. It writes an empty line and moves on.

Based on my experience following athletics competitions over twelve years, most data discrepancies do not come from deliberate fraud. They come from an administrative gap filled by guesswork, and guesswork is always cheaper than verification.

The crux is this: the system only verifies the validity of a result when that result breaks a record. For every other result, the system only checks whether anyone objected.

3. The Wind Ruler and the +2.0 Threshold Game

Athletics has a rule that sounds highly technical: a sprint or long jump result is only ratified as a record when the tailwind does not exceed +2.0 metres per second. That figure was fixed long ago and has barely changed.

The problem is that the rule speaks only to records, not to placings. An athlete who runs 9.95 seconds with a +2.4 wind still wins the race. That result still enters the personal record. It still appears on the news board. It simply is not counted as a record.

In practice, the gap between “not a record” and “not a performance” blurs very quickly in the news stream. I have counted cases where a wind-assisted result was headlined with the verb “set”, while the wind note sat at the bottom of the article, in a font size four times smaller than the headline.

There is a more complex layer: altitude. Mexico City’s track sits at roughly 2,240 metres above sea level. Thinner air reduces drag, helping sprint and jump events produce better marks. Nairobi sits at about 1,795 metres. Sestriere once staged a track meet above 2,000 metres. The same athlete, in the same condition, can run faster in these places without changing anything in training.

Wind and altitude correction models exist in academic literature, but they almost never appear in the mainstream news stream. A 9.90 at 2,000 metres altitude with a +1.9 wind and a 9.90 at sea level with a 0.0 wind read the same on a scoreboard. They are not the same.

The strangest thing is never the error margin — it is the way people try to explain it. Asked about the difference in conditions, the standard reply is “everyone runs the same track”. That argument is correct at the level of a single competition and wrong at the level of comparison across seasons.

4. The Equipment Dividend and an Unclosed Argument

In January 2026, World Athletics published technical limits for competition shoes. The sole of a road racing shoe could not exceed 40mm; track shoes could not exceed 25mm. A shoe had to be available on the open market at least four months before competition use. The number and arrangement of rigid plates in the sole were restricted.

The rule came after a season in which two lines of carbon-plated, supercritical-foam shoes took almost every major marathon podium. Eliud Kipchoge had run 1 hour 59 minutes 40 seconds in a specially controlled event in Vienna in 2026 — that mark was not recognised as a world record because it did not meet competition conditions. By Berlin 2026 he ran 2 hours 01 minutes 09 seconds, and that was an official record. Kelvin Kiptum ran 2 hours 00 minutes 35 seconds in Chicago in 2026, then died in a road accident in Kenya in February 2026, leaving a record no one has touched.

Those numbers are real. The problem is that when they are compared with earlier eras, the equipment contribution is rarely subtracted. A shoe that saves a few percent of energy per stride, multiplied across more than forty thousand strides in a marathon, produces a gap measurable in minutes.

Meanwhile, on the track, another metric gets ignored: competition density. As more licensed development meets open up, the number of valid races per athlete per season rises, and the ranking system rewards that number. An athlete can climb the rankings by racing many small meets instead of running fast at one big one.

I often ask: where did this money come from, and what did it do along the way? In athletics, that question applies to prize money. A Diamond League leg pays roughly 10,000 dollars to the winner of an event. A low-tier European meet might pay a few hundred dollars plus travel. That gap is enough to create a points-hunting market, where schedules are designed around ranking software, not around opponents.

5. Quotas and the Arithmetic of Selection

Entry to an Olympics or a World Championships has two pathways. The first is hitting the qualifying standard inside a published window. The second is accumulating enough ranking points over twelve months.

The second pathway creates a structural problem. Ranking points depend on the tier of the meet, the athlete’s placing, and the results of opponents in the same race. For countries with great depth, that pressure becomes an internal contest in which the fourth-placed athlete may be stronger than another country’s champion.

In Jamaican and American men’s sprint events, national trials have produced cases where the entry list excluded athletes with the season’s best personal marks. In Kenya and Ethiopia, distance selection races carry similar density. An athlete who finishes fourth in that system stays out of the major championship, and on the world ranking that absence looks like a weak season.

What the ranking system cannot measure is the quality of the people who are not on the list.

There is a further layer rarely mentioned: the cost of hunting points. An athlete outside a country with a strong federation must cover flights, hotels and entry fees at every meet. With a few hundred dollars of prize money at a low-tier meet, the economics only close when the athlete has a personal sponsor or an athletics scholarship at a foreign university.

That is why I always check the funding structure before checking the performance. An athlete racing many small meets in one season is usually servicing debt, not accumulating points.

6. The Silence of a Clean File

The Athletics Integrity Unit was established in 2026 and headquartered in Monaco, operating separately from the federation’s governing body. It publishes sanctions, whereabouts violations, and abnormalities on athlete biological passports.

The biological passport tracks biological markers over time, allowing detection of an anomaly that a single test would miss. The whereabouts system requires athletes in the testing pool to file daily location information. Three missed tests within twelve months constitute a violation, even with no positive sample.

Christian Coleman was suspended over whereabouts failures, with the sanction adjusted across appeal levels, and he missed an Olympic Games. Blessing Okagbare received a multi-year ban from 2026 for prohibited substance violations. Alberto Salazar, coach of a major national training project, received a four-year ban in 2026.

Those cases were published. But most testing activity produces no published result at all. An athlete tested ten times a year and negative ten times generates no news. An athlete never tested also generates no news. Those two files look identical in the news stream, and they are completely different in informational terms.

I often say: safety is not about never being caught, it is about never leaving a trace. In athletics, the trace most easily overlooked is silence.

The absence of a negative signal in a file that has never been tested is not evidence of cleanliness. It is only evidence of a file without a source.

This is where probability modelling becomes useful. If the testing rate for a group of athletes sits at a low level, then the probability of detecting a violation in that group, assuming other conditions equal, is correspondingly lower. A negative result in that group carries less information than a negative result in a heavily tested group.

7. The Grey Zone of Eligibility

Between 2026 and 2026, world athletics went through several regulatory changes concerning eligibility for women’s events. Testosterone regulations for events from 400m to one mile were maintained after a Court of Arbitration for Sport ruling in 2026, with a measured threshold of 5 nanomoles per litre of blood.

Alongside that, nationality transfer rules require a waiting period before an athlete may represent a new country. After a national federation is suspended, some athletes are granted neutral status so they can compete under a neutral flag.

These rules create a group of athletes absent from the ranking system for administrative reasons, not sporting ones. When analysing an event, I always check whether the field is thinned by eligibility conditions. If it is, every conclusion about that event’s relative strength needs its confidence level lowered by one notch.

The most memorable lesson I learned on this subject came from a file with no one’s name in it. In 2026, I followed a logistics audit of a major international event and found that declared meal allocations far exceeded the number of people actually operating. That file contained not one athlete’s name, but it taught me that a file missing names is not an empty file. It only means someone chose not to fill in the last line.

8. Medal Reallocation and the Value of Time

One feature unique to athletics is the possibility of redistributing results. When an athlete is found in violation and their results annulled, the placings behind them move up, medals are re-awarded, and earlier seasons’ rankings are corrected.

The big reallocation wave of the 2010s concerned earlier Olympic Games and World Championships, when stored samples were retested with new methods and returned positive. Athletes received medals after their careers had ended. Some received them at reallocation ceremonies held at a different competition, years later.

In data terms, this creates a specific problem. Result tables stored by media outlets, sponsors and data platforms were frozen long ago. When results change, those holding the old version do not update. Two versions of the same competition coexist on the internet, and the version with more search traffic is usually the wrong one.

I once traced a personal sponsorship dossier priced on a silver medal at a World Championships. Four years later, that silver became gold. No clause in the contract anticipated this, and no sponsor volunteered extra payment. The commercial value of an athletics result is not priced on the final outcome. It is priced on the outcome known at the moment of signing.

9. The Reasonable Side of the Skeptics

There is a counter-argument I consider largely correct. Those who say most athletics results worldwide are honest have grounds to believe it. Every year thousands of meets are staged, hundreds of thousands of results recorded, and most of them are timed by compliant equipment, signed by named officials, stored in systems with meet codes.

Most of the sourcing gaps I record are not fabrications. They are real results from real meets, whose administrative trail was lost at some intermediate stage. A provincial development meet never sends its technical report to the national federation. A local data supplier closes down and its files are not transferred. A wind gauge runs out of battery mid-session and the operator writes down the final figure by hand.

The skeptics are also right on another point: questioning every number carries a cost. That cost is paid by small federations, not large ones. A rule mandating source verification at every level would remove most small meets from the system and indirectly narrow the path for athletes from countries without sports infrastructure.

That is why I do not write about fraud before writing about the gap. All I do is connect the dots — and count how many people deliberately draw them wrong. Most dots are not drawn wrong out of malice. They are drawn wrong because no one is paid to draw them right.

10. Who Pays for the Last Line

File 187 is still on my machine. That 10.31 is still circulating somewhere, perhaps in a junior ranking, perhaps in a recruitment brochure for a sports academy, perhaps in a set of odds. No one in that chain has an incentive to go back and fill in the last line, because that line generates no reads, no contracts, no revenue.

What is notable is how the matter ended. After I published my report on that data batch, the supplier sent an update. The update added a competition date field to 189 of the 214 files. The remaining twenty-five were marked “undetermined”. File 187 was among those twenty-five.

A system chose to write “undetermined” rather than write a wrong number. That is progress. But a number marked “undetermined” is still used by those who do not read the notes field.

The question I leave behind is not for the data supplier and not for the federation. It is for the person reading the results table the next morning: when a number has no signatory, no date, and no meet, then whose number is it?