The Discipline of the Number-Keeper: When Data Is Empty, Don't Fabricate

2:47 AM. My computer screen was still on, and the data file was still empty. It was a night in late March, when I sat in front of a swimming meet's statistics sheet that I had promised my editor I would deliver before 8 AM. Every column showed zero: no split times, no stroke rate, no underwater propulsion data. Only a single note was left behind by my assistant: "source unverifiable."

A newcomer to the trade would fill that gap immediately. They would take an average figure from a previous meet, assign it, and call it analysis. I once sat next to such a person at a press conference. He opened a spreadsheet, tapped in a few numbers, and confidently declared: "By my calculation, this athlete will improve by 1.2 seconds." Nobody asked where that number came from. Nobody cross-checked. And the next day's article still made the front page.

That night, I shut down my computer and called my editor. I said a sentence I still remember word for word: "I don't have three sources yet. I'm not writing it." It was one of the hardest decisions of my career, and also the correct one.

From a crumpled sheet to a sea of data

In 2026, I began my career at Thanh Nien Bao as a swimming reporter. Back then, "data" was just sheets of paper with hand-written finish times. The writer had to stand by the lane, click a stopwatch, and trust his own eyes. After every SEA Games, that sheet was crumpled in a jacket pocket, and that was the entire record of a whole year.

Twenty years later, everything changed at dizzying speed. A V-League football match now produces hundreds of thousands of data points per game: passes, pressures, distance covered, sprints above 25 km/h. A swimming meet can give us hundredths of a second for every 50 metres. The surface of the sports industry, at least on paper, is now a sea of numbers.

But there is a paradox few people talk about. The more data there is, the more temptation there is to fabricate. Because when real sources are empty, nobody has the patience to check fake ones. I call this the "dangerous void" of Vietnam's sports analytics industry.

When I was a senior analyst at Becamex Binh Duong in 2026, I had the chance to analyse all 26 rounds of the V-League. The results showed this club had an average PPDA of 8.4 — the lowest in the league, meaning they let opponents make only 8.4 passes on average before closing down. Binh Duong's xGA was 0.68 per match, and they kept 14 clean sheets. I wrote the piece "Binh Duong pressing — a game that doesn't need the ball" with 17 data charts. It drew more than 250,000 reads.

The thing I'm proudest of isn't the read count. It's that every single metric in those 17 charts was cross-checked against at least three sources: match footage, the coaching staff's GPS data, and an independent third-party statistics table. Three sources, no more no less, and every chart noted its exact timestamp.

Three sources, and the discipline of emptiness

There is an invisible pressure that no one sees, but every team fears. I named it: Binh Duong pressing. That concept isn't about any single individual. It's about structure: a midfield willing to run more, press earlier, and convert pressure into goals. But to prove it, I needed data. And to make the data trustworthy, I needed three sources.

In 2026, thanks to the reputation from that pressing piece, I was invited to work as a World Cup analyst in Russia for a television station. I built an xG prediction model from 180,000 shots across five European leagues, and predicted 14 of 16 knockout matches correctly. When I wrote that Croatia were "low on xG but efficient thanks to 23 sprints above 25 km/h per match", many fans criticised me as dry. I responded with a 5,000-word essay.

But that wasn't the most important thing I learned in Russia. The most important thing was a sentence I wrote to myself: xG isn't wrong, it's just that football is inherently irrational. After 2026, I learned to count the irrationality too. That is, I no longer treated the model as scripture. I treated it as a compass — but without it, we get lost.

In 2026, the pandemic shut down football worldwide. When the Bundesliga returned with 312 matches without spectators, I saw a giant laboratory that nobody wanted. I found that home advantage fell from 54% to 47%. Home teams' PPDA rose by 0.9, meaning away teams pressed higher when there was no crowd pressure. My piece "Empty stands, reversed fortunes" drew 180,000 reads and was consulted by a Premier League club.

When the stands are empty, every model collapses. I rebuild from the burned-out wreckage of data. That isn't a slogan. It's an exact description of the work: gathering the data fragments that remain, classifying what still stands, and rebuilding order from there.

In 2026, I applied the empty-stadium model to the Euros and the Tokyo Olympics. I wrote a 12-part series on Italy, showing that their midfield had covered a total of 4,200 km after the group stage, with a PPDA of 7.6. From the quarter-finals I dared to predict Italy would win it all, against the majority. At the Olympics, I used the same framework to assess Brazil U23 and got 8 of their 10 most important players right.

In swimming, the work is no different. When analysing Nguyen Thi Anh Vien or Nguyen Huy Hoang, I never look only at the final result. I look at every 50-metre split, at how they handle turns, at tempo and distance per stroke. An athlete can finish in the same time, but the split structure tells a completely different story. It is that structure which predicts the next swim.

That is why I never fill a number into an empty cell. No split time means no pacing analysis. No stroke rate means no efficiency analysis. No three sources means no article.

The transfer market: where belief gets priced

During the transfer window, noise drowns out signal. Every day brings hundreds of rumours, and most aren't based on any verifiable source. Fans drown in rumours, and writers drown with them. The only way I've found to keep my head above water is to rank rumours by evidence: follow the money, follow contract structure, follow the agent's movements. A release clause, a sell-on clause, a salary figure — that is the real story.

The transfer market is the only place where people pay for expectation, not for the present. A player who hasn't played 50 top-flight matches can be valued at 100 million euros. That is not a calculated gamble. It's a bare gamble dressed in the clothing of analytics.

And here is where I return to that 2:47 AM. When I have no split time for a swimmer, I don't fabricate. When I have no source for a transfer rumour, I don't push it out. The principle is identical, only the arena differs.

Correlation is not causation

There is a trap that sports data analysts fall into most easily, and it is far subtler than fabricating numbers. It is confusing correlation with causation.

A team that presses a lot tends to win a lot. But that doesn't mean pressing creates wins. Perhaps that team has better players, a bigger budget, or simply an easy run of fixtures. If I apply a Premier League model straight onto the V-League without cross-checking at least three domestic sources — the fixture list, referee form, pitch conditions — then I am selling readers a false prophecy.

I have seen imported models applied straight onto Vietnamese football. They look beautiful on paper. But they don't know that a V-League pitch in August can be entirely different from a Premier League pitch in December. They don't know that a match can be broken apart by factors data cannot measure.

Numbers don't lie, but people always find ways to lie with numbers. The one who lies with numbers isn't necessarily a fabricator. Sometimes the one who lies with numbers is simply someone too confident in a model, forgetting that the model was built somewhere else, for a different football culture.

In swimming, this trap is even more dangerous. An athlete improves after changing coach — people rush to conclude the new coach is more talented. But perhaps that athlete has just passed puberty, or has just recovered from a shoulder injury. The real causation lies somewhere the model can't reach.

Leave it empty, when it needs to be empty

That 2:47 AM, when I shut my computer and told my editor I didn't have three sources yet, I didn't think it was a heroic act. I thought it was simply the most basic discipline of the trade. But the most basic discipline is often the hardest to keep.

Vietnam's sports analytics industry sits at exactly the intersection I once passed through: enough data to do the right thing, and enough temptation to do the fast thing. The question I leave for young writers isn't "how much data do you have", but "when the data is empty, do you dare to leave it empty".

The next round of Vietnamese football will be decided by those who can answer that question.