The Data Void in V.League: When the Spreadsheet Is Empty, the Model Must Stay Silent

On Tuesday night I reopened the transfer-market tracking file I am responsible for. Fourteen rows, seven columns. The market-value column had numbers. The transfer-fee column had numbers. Three columns were entirely blank: minutes played in the domestic league, positional defensive metrics, and pressing data. A colleague messaged me: "Just decide, which team is stronger — we need it tonight." I replied that I did not know, and that was the most expensive answer I gave all week.

Seven years ago I would not have answered that way. In 2026, as a journalism student, I built a World Cup prediction model from xG and xA across five European top leagues over three consecutive seasons. The model gave Germany a 78% chance of reaching the semi-finals. Germany lost 0-2 to South Korea in their final group match and went out there and then. The model correctly picked 12 of the 16 knockout-stage teams, and it was wrong exactly on the team I believed in most. When the model fails, the data starts telling the truth.

Vietnam's top domestic league operates with 14 clubs in the annual season, and since 2026 VAR has appeared in a portion of matches. After each round, what reaches the audience is goals, cards, rounded possession figures and the table. What never reaches the audience is event data with coordinates, expected-goals metrics, passes allowed per defensive action, and running distance split by half. A football ecosystem can survive for years inside that gap. It cannot conduct serious analysis inside it.

Based on my experience watching these matches, from the stands to replays reviewed several times over, I see the same argument repeat after every round: two people watch the same match, reach opposite conclusions, and neither is wrong — because neither has anything to verify against. That kind of argument does not produce knowledge. It produces heat.

In January 2026, Vietnam beat Thailand 5-3 on aggregate in the ASEAN Cup final. Nguyen Xuan Son topped the tournament's scoring charts and broke his leg in the closing minutes of the second leg. That was a data story abandoned at the exact moment it became valuable: that run of goals was never explained by shot location, chance quality, or touches inside the box. The press called it form. I call it an uncollected dataset.

Three layers of data are missing in Vietnamese football, and missing systematically. The first is coordinate-based event data: every pass, tackle and shot has a start and end location. Without it, any claim about controlling a match is a feeling. The second is pressing metrics. The third is physical data, specifically distance covered and sprint counts. These three layers are not decoration for a match report. They are the minimum condition that makes a tactical judgement falsifiable.

Take the most recent example I could actually measure. In the 2026-19 season, the home-win rate in the Bundesliga was 44.2%. When stadiums emptied during the pandemic, that figure fell to 36.7%, and average goals per match dropped from 3.1 to 2.8. No law changed, no schedule changed, no player quality changed. Only the crowd disappeared. Home advantage is not sacred ground; it is a variable that was frozen in place. Anyone who uses the phrase "fortress" to explain a home win in the V.League should ask: if the stands fell silent, would that rate still hold?

The second example comes from Euro 2026. Before the quarter-final between Italy and Belgium, I analysed pressing data: Italy pressed with an average PPDA of 8.2 — meaning opponents were allowed only 8.2 passes before an intervention — while Belgium covered 17% less ground than in their own earlier matches. I concluded Italy would control the game. Italy won 2-1. That was the first time a model of mine, with fitness context attached, correctly predicted a meaningful passage of play, and I learned that what decided it was not the PPDA number itself but my knowing the conditions in which it had been measured. PPDA is the signature; running distance is the confession.

In the V.League, neither exists publicly. So when someone says a club has "clearly improved its pressing", I do not hear data. I hear that person's memory of the last three matches they happened to watch. Memory is not wrong, but memory has enormous variance.

VAR, introduced in 2026, is an opportunity wasted in a different way. Every time a referee leaves the monitor to review an incident, there is a timestamp, a situation, a decision. If those traces were stored as a dataset, we would know which teams are genuinely disadvantaged, in which type of incident, and at what stage of the match. Instead, refereeing disputes are still handled with clipped footage and tone of voice. Data does not get emotional, but it remembers everything the press forgets.

The transfer market is the same, and it is where I work every day. In 2026 I followed Enzo Fernandez's move from Benfica to Chelsea at a fee of 121 million euros. I used World Cup data — 82% pass accuracy, 14 successful tackles — to build a valuation report. But the deal also depended on intermediaries, on the structure of payment terms, and on the buyer's urgency. Data cannot see those things. A transfer does not pick the best player; it picks the player you mis-measure the least.

There is a dark side effect of the digitisation of sport that few people in Vietnam state plainly: the very event datasets collected for analysis are also raw material sold directly to betting companies. When a league publishes no open data, private data investors hold the entire information advantage. The data gap in the domestic game does not get filled by journalism. It gets filled by the betting market.

I was born in France, I work in Shenzhen, and I read Vietnamese football through both lenses. What I see most clearly is European models imported wholesale and applied to Asian data without anyone checking which coefficients still hold. A model built on 90-minute rhythms in the Premier League is not automatically right in a league with different fixture density, different rest periods and different pitch conditions. That is not the model's fault. It is the fault of the person using a model without empirical verification.

The counter-intuitive angle sits here. Most fans believe the problem with Vietnamese football is a lack of data, so all we need to do is feed data in and everything becomes clear. I do not believe that. Correlation is not causation, and a denser dataset does not automatically produce a truer conclusion — it only produces more ways of presenting a wrong one.

The second risk is more serious. An empty table is not neutral. It is a bias engine, because people always fill a gap with whatever is easiest to recall, most emotional, and most recent. That is why notions such as "bogey team", "big-club mentality" or "head-to-head tradition" persist so stubbornly in conversations about the V.League. They survive not because they are true. They survive because we have never done the work of measuring them to prove them false.

And this is the part I want to say to myself rather than to anyone else: a model that returns an empty result has still produced a result. When my file has seven columns and three are blank, the right answer is not to infer from the remaining four. The right answer is to record that three are missing, note the date of the check, and tell the person asking that I will not commit. That is discipline, not timidity. I trust variance more than I trust champions.

Looking ahead to the next round, the signal I am tracking is not in the table. It is in whether any club starts publishing its own match data. It is in whether the organisers store VAR intervention logs as an open dataset. It is in whether, next season, anyone writes about a home win without using the word "fortress".

If all three answers are no, Vietnamese football will keep producing plenty of opinions and very little evidence. And fans will keep being asked to believe conclusions that no one is able to contradict. Can a football ecosystem progress that way — or can it only progress by learning to say: "I do not yet have the data to answer"?