The Data Gatekeeper: When the Spreadsheet Comes Back Empty

Hồ Minh — data journalist

The Data Gatekeeper: When the Spreadsheet Comes Back Empty

A Saigon night in November. I reopen the spreadsheet after a match I had followed all week, and the data column comes back a silent grey. No xG. No zonal heat map. No touches inside the box. The screen I had squinted at for nearly a decade suddenly turns into a blank sheet of paper — and that blank sheet is the hardest test in the trade.

In this job, everyone eventually receives an empty analysis. What matters is not the empty table but what you do when it is empty. A very human instinct pushes us to fill the grey cells with guesswork: this team has "fighting spirit", that player has "big-game nerve", that backline "throws itself into tackles". Those lines sound like football, but they are not data. My model does not cry and does not celebrate, but after every match it owes me a lesson. Tonight's lesson is that staying silent at the right moment is itself a conclusion.

Context: data born from zero

To understand why an empty table matters, you have to remember that Vietnamese football data was never a luxury good. The first xG table I wrote by hand on a bus, back when nobody called it data. In 2026, at 35, I retyped every passage of play from a V.League season for 14 clubs, just to build the league's first expected-goals model. No data provider poured anything into my machine. Everything had to be scraped from video, from match reports, and from my own memory.

Vietnamese football back then had almost no open data system at all. To know whether a team pressed high or low, I had to count myself how many passes the opponent made before being cut out. To know whether a striker was genuinely valuable, I had to convert every shot into a probability. Learning to build data from zero shaped me into a different person: someone who distrusts every statement that carries no measurement.

It was during that work that I met Phan Văn Đức — then just 20, a winger at SLNA. His xG per match reached 0.48, above the average of foreign strikers in the league, even though he scored only 5 goals all season. I wrote a piece predicting he would become a national-team pillar within three years. Many people mocked me for "daydreaming on numbers". In 2026, Phan Văn Đức scored the decisive goal at the AFF Cup. The lesson is not that I guessed right. The lesson is that I only dared say it because the data was real — not because I liked him.

That is also why I always set aside a paragraph in every analysis to admit my own limits. What is the sample size. What were the pitch conditions. Which variables the model still cannot measure. Readers deserve to know what ground I am standing on.

The null-handling principle

An empty analysis is not a failure. It is a datum. The problem only arises when the writer treats that emptiness as a shame to hide and fills it with reasoning that has no verification coefficient. I call my principle "null handling": when there is no data, say plainly that there is none, and never turn a gap into a fake conclusion. This principle is not attractive, does not trend, but it is the only thing that keeps this trade trustworthy.

There are two kinds of football writers. The first starts from a story and then hunts for numbers to illustrate it. The second starts from a number and lets the story emerge on its own. I belong to the second. The world saw Croatia as an underdog; I saw them as a string of coefficients no one had dared to exploit. At the 2026 World Cup, I applied the PPDA model — a measure of adjusted pressing intensity — to the big teams. Croatia under Zlatko Dalić recorded a PPDA of just 7.9 against Argentina, lower than that of a possession-branded side like Spain, yet they applied direct pressure extremely effectively. I wrote a long piece predicting Croatia would reach the final. A colleague laughed in my face because "nobody rates Croatia". When they beat Argentina, then Russia, then England in turn, my article was shared furiously.

But if that day I had only an empty table instead of the number 7.9, I would have written nothing. I would rather stay silent than bet on a feeling. That is the line between an analyst and a well-dressed fan.

That line becomes even clearer when I look at the transfer market. Every window, noise drowns the signal. Names are screamed on social media, contracts are rumoured with astronomical figures, and very few people bother to read the structure of the release clause or the real wage bill. The transfer market is a game for those who see far, not those who see much — value always arrives after patience. A small club signing a loan deal with an obligation to buy is not reinforcing its squad; in all likelihood it is tying its own hands to a payment it will no longer control two years later. The financial plan is upended at exactly the moment the club needs stability most.

The same logic applies to the arguments on the pitch. People believe video technology will extinguish every dispute over referees. But looking at the leagues I follow, the opposite is true: the dispute does not vanish, it simply moves. From the moment on the pitch, it shifts to the review room, to the definition of a grey zone in the law, to the question of who gets to redraw the line of a heel. A disallowed goal does not make fans less angry; it only changes whom they are angry at. Technology cannot manufacture clarity where the law itself is vague, and a good analytical model has to see that limit instead of pretending every passage of play can be reduced to a right-or-wrong bit.

I learned much during a period many believed was dead. In March 2026, the major leagues postponed en masse because of the pandemic. With no matches to analyse, many colleagues turned to entertainment writing. I did not sit still. I spent six months digging back through V.League data from 2026 to 2026. In 2026 the stands were empty, but every ball still fell into a cell of my model, and I understood that data never befriends a pandemic. In that pile of data, I found a rule that made me sit down: clubs that changed their chairman mid-season won 23% fewer of their next five matches. The cause was not on the pitch but in the boardroom — a governance shake-up drags personnel shake-ups behind it, and players sense it before the table shows it.

After my five-part retrospective series was published, an executive of a club called to thank me. He said it helped him avoid sacking his head coach at a sensitive moment. Data is not only for predicting who wins the title. It exists to keep people from rushing into decisions.

Contrarian: too much data is more dangerous than too little

Here I must argue against myself. People tend to believe that the more data they have, the less they err. That is not true. The most dangerous thing in the data trade is not too little data but too much data without context. A model run on a three-match sample can produce a very beautiful trend line, and that trend line can be entirely meaningless. I have seen tables presented as truth merely because they carried many digits after the decimal point.

Correlation is not causation. That 23% figure does not mean that whenever a chairman changes, the team will lose. It only says that, in the sample I hold, the two events travel together more often than chance. To turn it into causation, I would still have to exclude fixtures, injuries, home-and-away form, and all the qualitative variables that cannot be measured. And because I was born into a football culture where data is still young, I am forced to be humbler than those who work where every source is available.

The paradox is that precisely because I understand the model's limits, I dare to trust it elsewhere. I do not believe the coach; I believe the model. But I listen to the coach to fix the model. One coach told me he deliberately chose a deep defensive line to bait the opponent, and my model called it passive. That conversation forced me to add a new variable. New data always has the right to beat old data.

This also holds true for torn knees. A player returning from injury is usually greeted by a round of applause from the media, and short-term metrics can make him look as if he never left the pitch. But my model here does not measure running speed; it measures hesitation. People talk about the body, while the hardest thing to heal is the mind — the fear of not daring to go into a tackle, of not daring to change direction. Pushing a player back earlier than the data allows is destroying the second phase of his career, and no recovery chart can repair that.

What remains

Back to the empty spreadsheet of my Saigon night. I save it, I do not delete it. Because at some point I will have enough data to fill it, and by then I want to compare the new conclusion with the original emptiness. The data gatekeeper is not someone who rejects every story. He only rejects telling a story with numbers he has not verified.

Next season there will again be underrated teams. There will again be inflated contracts. There will again be matches where an anomalous metric whispers something before the score does. My job, every morning, is to keep my pen honest to the number — even when the humblest number is zero.