The Art of Recounting: When Sports Data Tables Are Empty

On the stands of My Dinh Stadium, on an afternoon in June 2026, I clicked my stopwatch for the third baton exchange of the men's 4x400m relay at the national youth athletics championships. Hanoi finished second, exactly 0.8 seconds behind the winners. That whole evening I rewound the footage, matched frame by frame against my handwritten notes, and found where the trajectory broke: the receiving runner started 2.1 metres earlier than the standard, forcing the outgoing runner to decelerate over roughly half a stride. Nobody fell, nobody collided, there was no spectacular moment worth cutting into a highlight clip. The error sat in a distance the naked eye cannot see.

To someone holding a stopwatch, 0.8 seconds is where the trajectory breaks. And that was the first time I understood that writing about sport does not begin with inspiration, but with a ruled sheet of paper.

That summer I wrote a long analysis on my personal blog, with a hand-counted data table: split times for each 100-metre leg of four athletes, intended versus actual exchange distances, speed differentials over the final 20 metres. A sports editor shared it. The comment section split into two camps. One said Hanoi lost on psychology. The other said the officials got it wrong. Neither camp mentioned the 2.1 metres.

A year later, at the 2026 World Cup in Russia, a new sports outlet invited me to contribute. I picked Russia versus Spain in the round of 16, not because the match was famous, but because I was curious about a detail the official statistics did not mention. The statistics recorded that Russia had 12 corners. They did not record how many times Russia repeated the same near-post header routine.

So I sat and counted. Seven times. Two of those created genuinely dangerous chances. Spain dominated possession, completing roughly twice as many passes as their opponents, yet in extra time their defence collapsed at exactly the point they had been warned about for 120 minutes. The piece ran under the headline "Russia were not lucky, they repeated the tactic seven times" and passed 50,000 reads. But what stayed with me was not the readership. What stayed with me was the feeling of realising that an official dataset can be complete in quantity while empty in meaning.

This is where I want to linger, because it is the boundary between analysis and decoration.

A statistics table exists to answer the question "how many". It was never designed to answer the question "how". When a player takes 12 corners, the statistics table has done its job. When a team takes 12 corners and seven of them follow the same trajectory, the statistics table has missed the entire story. Repetition is the one kind of data official statistics never count, and it happens to be the kind that says the most about a team.

In esports the gap is wider still. A clean KDA can conceal a player who only enters fights after his teammates have already won them. A high creep score per minute can simply reflect a team that is losing and funnelling resources into one person. Gold-to-damage conversion, fight participation rate, opening-fight win rate — each carries its own noise threshold, and that threshold shifts with every patch. That is why I always keep a notebook beside the screen.

When a team repeats the same routine seven times, they are not hoping for luck, they have engraved the tactic into muscle. I have tested that sentence many times, across many disciplines. In 2026, when competitions paused because of the pandemic, I built a small database on 40 Vietnamese track and field athletes: injury recovery time, competition frequency, peak age. A sports medicine doctoral student helped me construct a "record replicability" index. Early that year I wrote in my notebook that Nguyen Thi Oanh had a high probability of breaking the national 3000m steeplechase record. The record came, in 10:05.23. I kept every source and every calculation, not to show off, but so that if I was wrong I could still find the step where I went wrong.

A national record is not born in the final second; it is gathered across thousands of recovery sessions.

That same year, at the Tokyo Olympics, I wrote about the men's 1500m final. Norway's Jakob Ingebrigtsen covered the last 200 metres in 24.7 seconds, 1.2 seconds faster than the runner-up. I contacted an American coach to ask about the gear-change technique and the inside-lane starting position. I drew a speed chart to show how the banked curve reduces centrifugal force. That piece earned me an invitation to work as a documentary screenwriter. But what let me write it was not scene-building skill. It was a hand-counted data table nobody else bothered to count.

Here I have to say the thing few people in this trade want to hear.

The biggest risk in sports analysis today lies elsewhere: data that looks complete. An empty statistics table is obvious to everyone. A statistics table copied from official sources, with every row and column filled and clean formatting, is checked by almost no one. And when it is fed into an automated pipeline, the output can be fluent, confident, perfectly structured, and contain not a single fact.

I once witnessed exactly that in an esports analytics project. The input source failed to load. The system raised no error. It went on producing a nine-section report, full of terminology, full of conclusions, with not one line connected to a real match. A reader of that report would never know they were reading a decorated blank page.

To me, a null result has never been a clean result. When I cannot count, I write four words in the notebook: insufficient data. That is an honest answer, and in this trade it is worth more than any prediction.

Data analysts are moving into the dressing room. They bring charts, models, probabilities. Most of them have never sat in the stands at seven in the evening, clicked a stopwatch for a baton exchange, and felt that an athlete's breathing rhythm does not match the rhythm of a spreadsheet. Their conclusions are usually correct mathematically and out of step in time.

I begin with a hand-counted data table, because memory does not know how to make room for error. Every match is a wager that can be counted. You only have to be willing to watch.

What I want to leave behind is not advice but a habit: every time you read a statistics table, ask yourself where it was counted, by whom, and when. If the answer is "unclear", then that table is not yet data. It is only the appearance of data.