Five Data Gaps That Make Athletics Marks Misread

Five Data Gaps That Make Athletics Marks Misread

On my screen there are seven columns. The fifth is empty. That is the 200m split column — the only thing that tells you whether a 400m runner distributed her effort or simply ran out of hope. The other six are full: bib number, name, mark, start time, weather, referee. A file that looks complete. But with the split column blank, I lose the ability to tell two identical marks apart: the athlete who closed with the fastest final 100m of the field, and the athlete who collapsed after leading for 300m. Same result. Two opposite stories. One empty cell erased one of them.

I have turned down plenty of requests along the lines of “just tell me who is stronger”. I decline. Every number is a testimony. My only job is the interrogation. When the witness fails to appear, the verdict has to wait — not because I am overcautious, but because athletics is a sport where error is measured in hundredths of a second, and there an empty cell carries more weight than a paragraph of opinion.

Athletics runs on a chain of verification longer than almost any other sport. A mark is born on the track, passes through electronic timing, through the wind sensor, through the referee panel, through the national federation, and only then reaches the international ratification desk. Any link can snap. And when a link snaps, the information does not vanish — it merely becomes vague, and then gets filled in with feeling.

Based on my experience covering regional athletics meets, the gap between the roar of the crowd and the arrival of verified data is exactly where distortion is manufactured. The scoreboard flashes a mark within seconds. Screenshots spread within minutes. By the time the official results file is published with wind readings, splits and equipment declarations, the audience has shrunk by an order of magnitude. Most of the public has already formed a judgement before the data shows up.

For Vietnamese athletics, that lag is especially expensive. Names like Nguyễn Thị Oanh or Bùi Thị Thu Hà are invoked constantly, but most conversation stops at the medal and never reaches the metrics. A medal is an outcome. Metrics are capability. Between the two sits an entire data system that not everyone has the means to access, and not every organising committee has the means to operate. A wind gauge placed in the wrong position, a referee panel short of split-timers, a report that omits track temperature — those small details add up to a distorted picture broadcast as fact.

Five types of gaps recur in every athletics dataset I have opened, and each one is capable of reversing the conclusion.

The first is unrecorded environmental conditions. International rules allow a mark to count as a record only when assisting wind does not exceed 2.0 metres per second in sprints and horizontal jumps. Above that threshold, the mark is still published but removed from record lists. Altitude works the same way: in Bogotá at 2,640 metres or Mexico City at 2,240 metres, thin air reduces drag, and the advantage in sprint events can reach several tenths of a second. A 10.9-second run on a windswept coastal track is not the same class of performance as a 10.9 in an enclosed stadium. Without a wind column and an altitude column, we are comparing things that cannot be compared.

The second is the equipment dividend. Since 2026, the world governing body has capped sole thickness: a maximum of 40 millimetres for road and marathon shoes, and 25 millimetres for track spikes, alongside a restriction to a single rigid plate. In 2026 an additional clause required shoes to have been available on the retail market for at least four months before competition use. These figures go far beyond minor technical detail. They determine who is permitted to run fast.

The clearest example sits in the marathon. On 12 October 2026 in Vienna, Eliud Kipchoge ran 1:59:40. The performance was not recognised as a world record because of rotating pacers and hand-delivered drinks. Four years later, on 8 October 2026 in Chicago, Kelvin Kiptum ran 2:00:35 and was ratified as the world record. The two marks sit 55 seconds apart, but the difference between them lies in validity, not speed. Anyone quoting the number while ignoring the ratification paper is quoting a falsehood.

The third is a sample that is too small. One fast race does not constitute a level. Cross-referencing multi-season datasets, I find that an athlete whose personal best is 1.5 seconds quicker than the average of her other five races in the same season is usually not someone who has stepped up a tier — she is someone who had one perfect day. A perfect day is data. It is just not data about capability.

The fourth is unratified training marks. Hand timing carries a standard error of roughly 0.24 seconds over 100m, enough to turn a decent athlete into a phenomenon. Satellite watches err by 1 to 2 percent, and over 400m, 2 percent is nearly a full second. Downhill stretches, treadmill runs, artificial wind assistance — all can manufacture a beautiful number with no wind sensor, no referee and no official report. History does not repeat, but it keeps humming the same tune: every decade produces a few “training phenomena” that quietly vanish from every record book.

The fifth is missing splits. In the 400m, the 200m split is the whole story. In the 800m, the 400m split decides it. In the marathon, first half versus second half reveals whether an athlete is racing with her head or with her legs. Someone who closes faster than she opened is improving. Someone who fades by two minutes is emptying out. Identical finishing times. Two completely different diagnoses, and both disappear when nobody records splits.

This is where a counter-intuitive point emerges. Missing data is usually treated as neutral data, as though a blank cell simply means information is not yet available. That is wrong. Missing data is missing systematically, and the system carries bias. Races run in rain, into headwinds, on poor surfaces are the first things insiders forget. Beautiful races are the first things shared. The public dataset we read is therefore not a record of athletic capability — it is a collection curated by excitement.

The 40-millimetre sole limit and the 2.0 metres per second wind threshold are administrative conventions, written to manage a sport, not physical laws waiting to be discovered. Arguments about them are arguments about rules. That is entirely legitimate, but it should be named correctly. When someone says a record has been “diluted by shoes”, they are talking about a line in a regulations document, not about a biological truth.

Data never argues. It simply exposes what happened. And sometimes the truest part of the story is: we do not yet know. A file with an empty column, honestly presented, is more trustworthy than a file filled in with adjectives. Mispronounce one syllable and you rebuild an entire reputation — I learned that after a commentary shift in which I got a player's name wrong three times in the first half. Since then, every piece I write opens with a checklist and closes with a single question: which column is still blank?

The variable worth tracking next season is not the medal count, nor the national record. It is the quality of the results file published after each meet: are there splits, is there a wind reading, is there an equipment declaration. A track and field programme that publishes all three moves closer to regional standards automatically, because it is forced to look directly at where it is weak. When the world stands still, read the old charts again — and check which column is still empty.