The Result Sheet That Came Back Blank: Auditing the Athletics Data Chain Where N/A Does Not Mean Safe

The result sheet had eleven columns. Nine of them read "N/A". The other two carried only field names — athlete, event — with no value in front of them. At first glance it was an unfinished form. For me it was a finding.

Since I was eighteen I have read athletics results as claims, not verdicts. Every mark is a sentence somebody wrote, and behind every sentence there is a writer: a timing contractor, a race referee, a federation's records clerk. When a file comes back blank, the question is not what is missing from it. The question is which layer it went blank at, and how much trust had already been stacked on top of that layer.

Working from a US timezone, the federation PDF and the results feed are usually all I have. Both arrive formatted and authoritative. That is exactly why I keep a standing checklist of failure points, and why a null return gets filed as "absent evidence" rather than "bad data". The two categories produce different journalism: one gets reported, the other gets compressed into silence.

A blank column means one of two things: the document never arrived, or it arrived and was never read. Either way every downstream conclusion is void — but "void" and "risk-free" are not the same word.

How a mark is manufactured: five layers

No number in athletics arrives alone. When you see "10.12" on a screen, it is the last link in a chain of five documents, and any link can drop something.

The first layer is the entry list. Start lists are never stable. Late withdrawals, scratches, additions — all routine. What is not routine is the gap between the published list and the real one, which rarely makes it into a PDF. I once counted a national meet whose entry list was updated three times, each time under the same file name. The reporter who downloaded version one and stopped there wrote an entire piece off the wrong heat sheet.

The second layer is timing. Fully automatic timing reads to 1/1000th through a photo-finish camera and publishes to 1/100th. Hand timing is the weakest link in the chain: it is not valid for records at all, and where it has to be used as an estimate, sprints take roughly a 0.24-second addition, rounded upward. That 0.24 is not magic — about 0.14 seconds of reaction time plus about 0.10 seconds of human response. Any report that carries a hand time forward without the correction has just made an athlete faster than he was.

The third layer is wind. Sprint and horizontal-jump marks count for records only with a tailwind of +2.0 m/s or less. In the 100m the gauge reads for roughly ten seconds from the gun; in the long jump and triple jump, for roughly five seconds from take-off. A mark that looks entirely legitimate can be void because of what a gauge recorded inside a five-second window. Where the wind column is empty, the mark is neither valid nor invalid — it is unclassified.

The fourth layer is the referee's report. Lane infringement, baton exchange outside the zone, false starts. Since the zero-tolerance rule took effect in 2026, a single false start is automatic disqualification. The 4x100m exchange zone moved from 20 to 30 metres from the 2026 season, and that change landed directly on teams whose preparation was built around the old marks. Without the referee's report you cannot know whether the time in front of you was amended after a disqualification.

The fifth layer is publication. This is where most things fall. DNF, DNS and DQ reclassifications, record ratification, corrections added later. During the 2026 pandemic hiatus I digitised historical Indian athletics records. The official time in the Delhi 2026 Commonwealth Games women's 4x400m relay contained a timing error. I cross-referenced twelve sources and interviewed two officials over three months before publishing. The Athletics Federation of India dismissed it, then quietly corrected it. The national record had a hole you could drive a truck through, and anyone who never looked for it never believed it existed.

That is why a file that returns completely empty is not a fifth-layer problem. It is a zeroth-layer problem. The input never arrived.

Four known failure points, and how each gets checked

First, hand-timed meets. Still common at district and state level. The check is mechanical: does the time carry one decimal place or two? One decimal almost always means a human thumb. I file such a result as a signal, never as a performance, and never compare it to a national-level mark.

Second, missing wind readings. Indoor meets, or outdoor meets with a dead gauge. The dangerous habit here is the analyst assuming the wind was "probably normal". That assumption never gets written down and always ends up inside the conclusion.

Third, late entry-list edits. Check the gap between the file's publication timestamp and the start of the meet. If the file changed hours before the start, at least one name in it will not be in the stadium.

Fourth, DNF/DNS/DQ reclassification. This is the quietest correction in the sport. An athlete who was fifth in the published result is fourth a week later, and nobody explains why. The audit revealed a decimal shift that rewrote the podium — but the interesting question was never who moved from third to second. It was what changed for the athlete. If the answer is "nothing but a line in a record book", the piece is a note, not a story.

My hardest self-rule: a correction is not news until you can say how the athlete's next morning differed — in funding, in qualification, or in a job.

Nine dimensions, and what N/A actually means in each

The worst mistake available to an analyst facing an all-null input is to convert emptiness into a universal negative: "no risks identified". That sentence is false. The true sentence is: "risk could not be assessed." Everything that makes a report readable rather than dangerous lives in the gap between those two.

Performance requires four things: the mark, the wind, the altitude, the equipment. Most Indian meets sit near sea level, so altitude correction is often assumed to be zero — but venues above roughly 1000m assist sprints and jumps and penalise endurance events. Where the venue is not even named, no correction applies. Where equipment is unreported, the 40mm road-shoe sole cap debate is simply absent from the analysis, because you do not know which model was on the foot.

Athlete condition requires a date of birth before the age curve can be applied: sprint peaks around 24–29, middle and long distance 26–31, throws 28–33, marathon extending past 35. The PB-progression anomaly test needs year-by-year data — a single-year gain of roughly three times the athlete's own historical annual average is the flag. Injury-history grading needs a record of two or more consecutive seasons lost. Without these, you cannot write one honest sentence about the athlete.

Competition structure has two qualification channels — the entry standard and World Ranking points — and they are checked differently, because one needs a date and a window and the other needs a whole season. On top of that sits national selection architecture; a US-style trials system can eliminate a reigning world record holder in a single race. Until nationality and federation are known, this entire dimension is unavailable.

Event landscape needs at least the season's top-10 list. Extrapolating a national pipeline from one athlete's good race is the error I see most: one fast heat, one imported coach, one funded junior do not make a trend.

Rules and anti-doping demand the most care. The absence of a doping red flag in a report is not evidence of a clean record — it is evidence of missing data. A check requires at least one signal: an abrupt performance leap, a whereabouts failure (three missed tests in twelve months constitute a violation), an association with a sanctioned coach or team doctor, or a low-testing jurisdiction. Samples are stored for ten years and medals get reallocated long after the fact. None of this can be evaluated on an empty input.

Team and training system means coach, school of thought, sports-science support, rehabilitation access. In the Indian context there is an extra layer: sports-quota jobs at Railways, Services and the public-sector undertakings are the real professional contract, and TOPS and Khelo India money are line items stacked on top. Until you know which system an athlete belongs to, you cannot say one concrete thing about their preparation.

Risk maps shift with discipline — hamstring and Achilles in sprints, hip and knee in hurdles, ankle and lumbar in jumps, shoulder and elbow in throws, stress fracture and plantar in distance. No event name, no map.

Public narrative has its clearest test case in Neeraj Chopra: lifetime best 89.94m, Tokyo 2026 gold at 87.58m, the Budapest 2026 world title at 88.17m, Paris 2026 silver at 89.45m. At seventeen I modelled a javelin throw and saw the future — release angle around 34 degrees, approach speed around 8.2 m/s, and the model said 90 metres was reachable within three years. That model still tells me the same thing today: the depth behind him is thin, while the discourse says the depth is forming. The gap is in perception, not in the data.

Industry transmission needs a shock origin — a competition, a brand, an athlete. With none of those, industry analysis collapses into general commentary, and general commentary does not get published under my method.

The pre-registered prediction, and what it is actually for

When I publish a model before a meet, I publish an interval, not an estimate. For the javelin it reads roughly like this: if approach speed lands between 8.2 and 8.4 and release angle stays inside 33–36 degrees, the expected range in the final is 84 to 89 metres, and the probability of clearing 90 in this sample is low. Three things are deliberately included: the inputs, the interval, and what the model cannot see — injury history, a gust, one fouled attempt.

The same discipline applies to an empty dataset. If my model returns null, my prediction is: no meaningful estimate can be produced from this input. And my failure condition is stated too: if the source document is restored and returns a title plus at least one information point, this prediction was wrong. It sounds absurd. It is the only way to stop emptiness from becoming an excuse.

What the spreadsheet didn't see was the one place in the pipeline where a document arrives and still does not arrive. A missing title, a missing source, a missing entity list and missing information points, all four at once, point to a failure in the reading step rather than in the article. Any published athletics item contains at least one of: an athlete, an event, a mark, a venue. All four vanishing together is itself a data point.

The contrarian angle: the problem is not missing data, it is the confidence stacked on it

Data scarcity is not a new disease in athletics. Hand times, missing wind, late entry edits — decades old, still here. What changed is that a thick layer of analysis now sits on top of that incomplete data. An incomplete result used to look incomplete. Now it lands in a dashboard with three colours and a percentage, and the reader assumes something was measured.

My profession builds that layer. My most useful work is scraping it back periodically to see what is underneath. When someone says "the data says", my first question is: which data, dated when, from which layer, verified by whom.

The second contrarian point is less comfortable. An empty output may be the most honest document in the whole chain. A file that says N/A is not lying. A file that looks full and is wrong is. Journalism treats honest emptiness as a embarrassment and filled-in guesswork as professional skill. That should be reversed.

Third, and this matters most to me: I could have written this whole report as a scandal. "Athletics data system fails" is a good headline and it would have earned clicks. I did not, because I have no evidence that any federation, competition or athlete was harmed by this failure. What I have is a broken process, not a damaged athlete. Blurring that line is easy, and it is the biggest professional offence available in this trade.

Fourth, the argument about analysts invading dressing rooms starts here. When the foundation is incomplete and the conclusion is complete, players and coaches are right to be suspicious. The actual rhythm of a meet — a hot afternoon, a delayed heat, a tightened hamstring — never enters the spreadsheet. An analyst who believes his model reaches the stadium is publishing his confidence, not his limits.

What to watch

This exercise has one practical output: let N/A stay N/A. When you compress a report into a summary, do not drop the nulls. When you write the headline, do not turn absent evidence into a clean record.

Four signals are worth tracking. Source recoverability: if the original article can be re-read and returns a title plus at least one information point, the full analysis can be re-run. Extraction integrity: compare the count of information points against article length; a long document returning zero points means the problem is in the reading layer, not the source. Domain-label reliability: the label "athletics" is attached even though no athletics entity was found, which makes the label itself suspect. Time-sensitivity backfill: without a publication date, neither the narrative cycle nor the peaking window can be anchored.

On the day all four go green, I will reopen this file. Until then, every N/A in this report stays exactly where it is — as a warning, not as a blank.

And the next time a dashboard looks full, ask one question: which layer sits under this number, and who wrote that layer's name?