The Athletics Record and the Data Nobody Bothers to Read

On the results sheet from an internal test in Nha Trang, the smallest line was the most important one: +2.3 m/s. Nobody in the stands noticed. They only saw 10.28 seconds, a handsome number, enough to make a few people pull out their phones and check whether it broke any record. I stood at the edge of the track, sheet in hand, and asked myself: if that small line were erased, would anyone notice?

I begin every conversation about athletics with the conditions that produced a performance, not the performance itself. An athlete who runs 100m in 10.28 seconds with a 2.3 m/s tailwind has crossed World Athletics' legal limit. The maximum is 2.0 m/s. The time is real, but it is not recognised as a record. There is nothing shady here, only a line of data left unread, and a crowd that accidentally misread the truth.

That is why I chose this job. Not to find the fastest runner, but to find the true meaning of speed.

Context: the regular season and a flood of times

We are in the middle of the regular season, when domestic and regional meets run back to back and the results sheets thicken every weekend. This is the most dangerous time for readers: too many numbers, too little time to verify. An athlete runs well once, and immediately someone calls it a turning point. A national record falls, and immediately someone calls it a golden generation.

I do not object to excitement. I object to excitement without close reading. Based on my experience tracking races over more than two decades, most mistakes in athletics analysis come not from a lack of data, but from reading data detached from its context.

An athletics performance is always the product of at least four variables: environmental conditions, equipment, race structure, and sample size. Ignore any one of them, and you are holding half the truth. A time can only be read beside the conditions that produced it; detached from them, it is just a handsome time.

The core: four layers of data behind one mark

The first layer is wind. It is the most underrated variable. World Athletics rules that marks in sprint events, long jump and triple jump are only recognised when the tailwind does not exceed 2.0 m/s. That threshold is not a formality. Over 100m, a 1 m/s tailwind can be worth roughly 0.05 to 0.1 seconds, enough to reshape an entire race. When I read a results sheet, my eyes find the wind line before the time line.

The second layer is altitude. Thinner air at high elevation reduces drag, producing marks that are hard to reproduce at sea level. At the 2026 Mexico City Olympics, at 2,240 metres of altitude, Bob Beamon long-jumped 8.90 metres, a record that stood for 23 years until Mike Powell jumped 8.95 metres in Tokyo in 2026. Beamon's mark is real, but without altitude beside it, it is a time read short. In Africa, many athletes train around Nairobi, some 1,795 metres up, which explains part of their endurance, not the whole story.

The third layer is equipment. This is the layer fans most often skip, and the one that has caused the most argument in the past half-decade. On 31 January 2026, World Athletics announced footwear regulations effective from 30 April 2026: road shoes limited to a 40 mm sole thickness and a single rigid plate; track spikes limited to 25 mm. The rules arrived after a boom in carbon-plated shoes, when long-distance records fell in quick succession. Reading a marathon record without knowing which shoe it was run in, and when relative to that threshold, is reading half the story.

Eliud Kipchoge ran 2:01:39 at the Berlin Marathon on 16 September 2026, an official world record. A year later, on 12 October 2026, he ran 1:59:40 in Vienna at the INEOS 1:59 Challenge. The later time is finer, but it is not recognised as a world record, because it was a staged event: with pacers, with a support formation, not an official race. Two times, two contexts, two entirely different meanings. A hasty reader will merge them into one.

The fourth layer is sample size. One race does not make a career, and one surge does not prove a fitness base. This is where I learned my most expensive lesson. In 2026, while working as a data consultant for a football site, I analysed Hanoi FC's first 15 matches and concluded they would win the title, based on chance quality rather than luck. The article was pilloried. Then they lifted the trophy. But what I took away was not that I was right. It was that I was right only because I had read all 15 matches, not one.

A year later, at the 2026 World Cup, I tallied Luka Modric covering 88.3 km across the tournament, but his sprint speed dropping 21% after the 70th minute. I wrote that Croatia would advance if they used their three substitutions to cover the distance. Colleagues laughed. Croatia reached the final. The lesson repeated: sample size and distribution over time tell the story, not a single moment.

When numbers speak, I simply listen. But I only listen after checking the conditions under which they were born.

There is one more layer I place in the core: race structure. Over 400m, people usually look only at the finishing time, while what decides the performance lies between 200m and 300m, where the athlete must hold rhythm while the muscles tire. Over 800m or 1500m, final-lap speed only means something beside the earlier laps. An athlete who finishes with the fastest last lap but whose middle laps are 3 seconds slower than their own best is telling me something about their fitness base, not just their speed.

The contrarian angle: correlation is not causation

There is a temptation every data analyst has met: seeing two times move together and concluding one caused the other. In athletics, that temptation takes the shape of training marks.

Every season I get a few messages like this: this athlete just ran 10.1 seconds in training, they are about to break the national record. I always answer with the same question: measured with what, where, was there an official timer, was there wind. Training marks are not recognised because they lack a chain of verification. One fast training run may be real, may be a mis-set watch, may be a tailwind, may be a rare explosion. There is no way to tell with a single sample.

This is why I set a rule for myself: never diagnose an athlete after one run. You need at least two, better a series, to separate signal from noise. At the 70th minute, the crowd sees collapse; I see a structure being rebuilt, but only when at least two independent metrics support that reading. Otherwise, I stay silent and wait for data.

People ask why I am silent; I am reading the words the track writes. When the data is empty, silence is not weakness. It is discipline.

What to watch in the next cycle

The regular season is long, and the results sheets will thicken further. I will not watch peak marks first. I will watch three things: the stability of middle laps, sprint speed over the final 20% of a race, and how many times an athlete repeats their best mark under equivalent conditions.

Distance never lies, we are simply not patient enough to listen. This season, I will spend more time listening, and less time declaring.