One Engagement Story, Nine Football Analysis Dimensions, and the Expensive N/A Lines

On October 1, a television series premiered on HBO/Max. Shortly before that, a newspaper in Karachi reported that a 44-year-old actress had been surprised with a proposal in Central Park, that the couple then appeared on The Tonight Show with Jimmy Fallon, and that a photograph of the ring had been taken in Barcelona. No team. No player. No contract, no league table, no referee, and not a single minute of football.

That record was still tagged “football”, then pushed through nine deep-analysis dimensions: tactics and technical assessment, club finance and the transfer market, results and the public-opinion cycle, league landscape, rules and governance, the dressing room, risk profile, media narrative and expectations, and industry transmission. Every table returned the same conclusion: insufficient football information, cannot be assessed.

I read that output twice. The first time, as a working journalist, I felt relieved: someone had refused to fabricate. The second time, I felt cold, because if this happened inside a real newsroom with real output targets, the letters N/A would very easily be replaced by three thousand words about the “fighting spirit” of a person who has never touched a ball.

The most notable thing in that report was not the nine empty tables. It was the note explaining why: the domain label had been assigned incorrectly. In this trade, admitting that the assignment was wrong from the root is the hardest thing to do. It wins no page views, no praise, only a blank space in the right place.

I have worked in sports journalism for 27 years, and I carry a stats sheet and a notebook with me. What I carry more than that is a cheap and difficult habit: only write when there is evidence. That mislabeled record reminded me that this habit belongs inside the system, not only inside individual conscience.

A single mislabeled record, standing alone, is almost harmless. What is worrying is that no record stands alone.

In a modern sports content pipeline, every item passes four stations: collection, tagging, analysis, publishing. The second station is the least inspected and the most decisive. People tag by headline, by source name, by keywords in the opening paragraph. A piece containing the words “player”, “pitch” or “league” is easily filed under sport. A piece about an evening in Central Park should have nothing that qualifies it as football, unless the tagger is chasing volume.

Volume pressure in Vietnamese newsrooms comes from two directions. Newsrooms need reads. Platforms need fresh content every day. Both prefer a broad label, because a broad label captures more items, while a narrow one loses them. In that race, label accuracy is the last thing left behind.

The cost does not sit in the article filed in the wrong place. The cost sits in the accumulated record.

Bad data does not do damage on the first row. It does damage on the ten-thousandth row, once a model has come to believe its dataset is clean.

If thirty entertainment records slip into a football dataset every month, that dataset holds seven hundred noisy rows after two years. A model that reads news, suggests topics, and ranks players by mentions will also learn names that have nothing to do with any match. The consequence does not detonate in a single day. It leaks slowly: a women’s player ranking carrying a name that does not belong to a pitch, a transfer item assigning the wrong club, a scouting report built on media fame instead of match data.

In women’s football, that noise is several times heavier. The database is thin, so each wrong row carries greater weight. A national women’s league may record only a few hundred events per round, while a men’s league of the same tier records several thousand. When the sample is small, noise stops being a speck of dust. It becomes part of the picture.

During the transfer window, the problem gets worse by a notch. Rumours already drown out signal, and fans are submerged in noise. My method each window is to rank information by evidence level: signed paperwork, release clauses, wage structure, agent movements, injury status. But if the base data layer is already contaminated by wrong labels, every filter behind it is useless. Nobody can filter clean a warehouse that was loaded with the wrong goods from the start.

In 2026, I sat down with 1,432 phases of play from one women’s national league season and built the statistical model myself, because no ready-made source existed for it. I remember the feeling of typing each phase: who passed, to whom, where on the pitch, under pressure or not. Some matches I watched three times just to confirm whether a through ball counted as a chance.

The result was a number that forced attention: a young forward at the Ho Chi Minh City women’s team converted 23% of her chances across 18 matches. The article drew 250,000 reads and brought a new audience to the league. What I kept afterwards was discipline: I knew exactly which phases I counted, which I discarded, and why.

Numbers do not lie – be patient enough to hear them tell the story of a girl running 90 minutes for hunger.

If I had allowed myself to guess that year, the 23% figure would carry no weight. It only carries weight because it came from a process that can be re-checked. That was the lesson I carried into 2026, when I was one of two Vietnamese women journalists accredited for the World Cup in Russia. During the France – Uruguay quarter-final, a male colleague cut me off live on air with a remark about looks. I answered with the data of the very player he was discussing: 11 successful dribbles, 4 chances created, 2 goals in 5 matches, and a clear performance lift when he drifted to the right flank in a 4-2-3-1.

The incident spread quickly on social media. What people remembered afterwards was how a position was defended, not the argument itself. Correct data has one property: it ends a debate faster than any slogan.

The mirror problem of mislabeling is what happens when data does not exist and people fill the gap with adjectives. Vietnamese women’s football lived in that condition for years. No footage, no match reports, no minutes played, no one keeping records. That gap was filled with “character”, “spirit”, “hunger”, “will”.

Those words are not meaningless. They are irresponsible when used in place of numbers, because they cannot be verified, compared, or tracked across seasons. A player praised for being “full of hunger” in seven consecutive articles still leaves nobody knowing how many kilometres she runs per match, how often she shoots, or how many passes she loses under pressure.

The pitch has no room for prejudice – only the ball, the tactics, and whoever dares to stand up.

Every phase of play is a piece of a puzzle – and every piece is a life waiting to be recognised.

What I propose is not a complex system, but three cheap checkpoints.

Checkpoint one: entity verification. Before a record enters a football dataset, the system must find at least one football entity inside it: a team name, a player name, a match name, a competition name, a match date. If none exists, return the original label. A story about an evening in Central Park stops right here.

Checkpoint two: separate “not enough information” from “no information”. These two states are merged into one in many processes, and that is the origin of most fabricated content. “Not enough information” means data exists but is thin and needs tracking. “No information” means the topic lies outside the domain. An engagement in New York belongs to the second type, and the correct handling is to stop.

Checkpoint three: log the audit trail. Every conclusion must carry a source and a timestamp, so that six months later someone else can trace where it came from. A number without a source is an opinion presented in digits.

The irony sits here. When a process returns nothing but N/A, the first reaction of most people in the industry is to treat it as failure. Failure of the analyst, of the model, of the system. Nobody praises an empty table.

But an empty table in this case is a correct result, and even the most valuable one. It reports a bad input, points precisely at the fault, and keeps the dataset clean for one more day. Compared with three thousand smoothly written words about a topic that does not exist, a table of N/A is worth many times more.

Keeping every sentence anchored to a real event – that is my trade, before it is content production.

There is another hollow nobody mentions. Mislabeling incidents are usually treated as technical errors. Most of them come from human choice: choosing to write instead of choosing to stop, choosing to file instead of choosing to ask again. An editor who understands the craft can perfectly well see that an entertainment item does not belong on a sports page. What pushes it forward is output pressure. Fixing a model is easy. Fixing a target is much harder.

And the reader? Readers do not see labels. They only see articles. Every time they read analysis built on junk data, their trust in sports numbers as a whole wears down a little more. That is the heaviest loss, because trust disappears fast and returns slowly.

On the evening of October 1, a television series premiered and had nothing to do with football. The story should have ended there.

What I want to leave behind is not a warning about machines, but an invitation. Next time you read a piece about women’s football and find it full of pretty adjectives and not a single number, ask yourself where the data is. If the answer is that there is none, the work to be done is not more writing. It is collection.

The lights go out, life goes on – I write about the women footballers who never leave the pitch even when there is no crowd.

Vietnamese women’s football does not lack good players. It lacks recorders. A league only matures when someone is willing to sit down, count each phase, and refuse to fill the blanks with words that sound pleasant to the ear. I do sports journalism to record the people who dare to dream in a world that has no place for them.