9:40 in the evening. The file arrived on schedule. Eleven columns, forty-two rows, and not a single cell contained a number.

I had watched that match from the first whistle to the fourth minute of stoppage time. I knew the home side shifted into a back four on 58 minutes. I knew the away side's right winger received the ball three times in the inside channel in the second half, and all three moves ended with a cross nobody attacked. But when the summary sheet opened, the xG column was empty. The PPDA column was empty. The column for ball recoveries in the final third was empty. Eleven columns, forty-two rows, every one of them returning a null value.

The person who sent it added a single line: "You can write it from what you saw."

I read that line three times. It sounded perfectly reasonable, and it was the most dangerous offer an analyst can receive in an evening, because it invited me to do the exact thing I had spent twenty years refusing to do: fill the gap with memory, then call memory data.

An empty sheet is still an intact sheet. It is simply telling you it has nothing to say yet. I closed the file, wrote "insufficient" in the log, and sent it back with a list of what was missing: shot counts by zone, per-shot xG values, the number of passes the home side allowed before each defensive action, and timestamps for every transition.

No article was published that night. It was the best decision of the week.

Context: a process with no room for convenience

To understand why a professional can sit still in front of a match she has watched in full, you have to look at the process.

Every analytical piece I hand to an editor runs through a fixed five-part frame. The opening must be an anomalous signal, not a greeting. The context section must state the data method and the intervening variables. The core is a chain of evidence ordered from raw to refined. The contrarian section must show where correlation has been misread as causation. The close must be a judgment pointing to the next round, not a summary.

Attached to that frame is a hard rule: every conclusion must trace back to a numbered information point in the original record. No number, no conclusion. That rule sounds dry until you realise it is the only thing separating analysis from guesswork.

My logging system has nine groups. Tactical and technical. Club finance and the transfer market. Results and the public-opinion cycle. League landscape and team positioning. Rules and compliance. Governance and the dressing room. Risk. Media narrative and expectations. Industry transmission. For each group I answer three questions: what is the conclusion, where is the evidence, and which assumption is holding the conclusion up.

When the input is empty, all nine groups return the same state. That state is not "weak". It is "not assessable". In my trade those two words are a valid result, not a surrender.

In 2026 I walked into a television sports department with a notebook and a rule set by my first editor: record the source of every number, even when you calculated it yourself. Thirty-five years later the notebook has become a spreadsheet, but the rule has not changed.

From the desk to the pitch: the same test applies to any match. And this is where Vietnamese football currently stands.

The volume of televised football in Vietnam is not small. V.League 1 runs a long season, domestic cups and youth competitions overlap it, and the national team plays qualifiers and regional tournaments. But the public data layer is far thinner than what European audiences are used to. You can find scorelines, minutes, scorers. You struggle to find a complete match-by-match xG series, a consistently published PPDA figure, or open position-tracking data.

The consequence is not that people stop analysing. The consequence is that people analyse with adjectives.

A team's "fighting spirit" is judged from players' faces after the final whistle. "Character" is judged from one moment in the 88th minute. "Honesty in pressing" is judged by how often a red shirt ran at a blue shirt, regardless of whether that run cut out a pass.

PPDA is not a measure of spirit; it is a measure of honesty in pressing. It counts the passes a team allows the opponent before making a defensive action. Lower means denser pressing. It does not care whether a player gritted his teeth. It only records what happened on the grass.

Core: four evidence chains and one test

In 2026 I took on a match in the Chinese top flight. The home side were reigning champions; the visitors were rising. My sheet returned xG of 1.2 for the home team and 2.3 for the away team.

Define the concept before using it, because I do not want anyone pretending to understand. xG, expected goals, is the probability that a shot becomes a goal, derived from historical data on hundreds of thousands of comparable shots by position, angle, shot type, and the number of defenders and goalkeepers in front. Add the xG of every shot in a match and you get a number describing the quality of chances a team created. xGA is that same number applied to chances conceded.

The bookmaker priced the home side as favourites at 1.85. I backed the visitors on the handicap, +0.5. A male colleague looked at my sheet and asked directly what a woman could possibly know about football.

The match finished 2-2. The +0.5 handicap won. I collected 40,000 yuan.

The lesson was not the winning bet. It was the order of operations. Define xG first. Run the numbers second. Compare against market price third. Only then speak. The phrase "I feel" did not appear in that record, because feelings have no unit of measurement.

From that day I built a standard template for every match: xG, xGA, shots by zone, possession, entries into the final third, and PPDA. That template has stayed with me.

In the summer of 2026, at the World Cup in Russia, I dissected the semi-final between France and Belgium with a single metric.

Belgium allowed 12.5 passes per defensive action. France: 8.2. In other words, France accepted conceding the ball and then won it back earlier, higher up the pitch, in fewer passes. Belgium pressed later, waiting for opponents to enter midfield before applying pressure.

The public called France pragmatic, boring, even cowardly. I wrote a piece titled "France were not cowardly, France were smart". A European magazine shared it and it passed 500,000 reads.

The match finished 1-0 to France. The only goal came on 51 minutes, Samuel Umtiti heading in Antoine Griezmann's corner. That is exactly the script a side that cedes territory and waits for high-quality chances tends to find, and it sits comfortably alongside Romelu Lukaku and his team-mates seeing more of the ball.

Numbers never lie; only the people reading them lie to themselves.

In 2026 global football froze. My data contract was cut by 60%. I had to rebuild a model from ten years of history, and I chose a single variable to test first: home advantage.

When the Bundesliga returned in May, the data showed home advantage down roughly 37% without crowds. I bet the model and won 12 of my first 15 positions.

Then I was wrong. For three rounds, teams played on the inertia of the previous season. By the fourth round, coaches had adjusted their approach to away games, because there were no crowds to absorb pressure. I refused to update the parameter, simply because I believed in the sample that had been winning. I lost the next four bets in a row.

The lesson was not that the model was wrong. The lesson was that parameters have a shelf life. Since then, every analysis I file ends with a section called "Assumptions and lag", stating what the model assumes and how many rounds behind my data is. The frame stays fixed. Parameters are updated on a schedule, not on inspiration.

That home-advantage shock taught me one thing: the only constant is change.

Euro 2026 was played with the pandemic still shaping the calendar. Roberto Mancini's Italy averaged around 60% possession, and most viewers described them as a controlling team. But a possession percentage says nothing about the quality of that control.

So I built a metric to answer the question: dangerous control. The calculation is simple: count the entries into the opponent's final 25 metres per 100 possession sequences. Italy led Europe at 18.2.

That number explains why Italy's story was never about owning the ball, but about moving it to where it can do damage. I backed Italy to win at 11/1 and collected 275,000 yuan. A European betting company subsequently hired me as a data consultant.

My meta-detection process now has three steps and always includes cross-checking. Step one: identify which metric the market is mispricing. Step two: check whether the sample is thick enough to support a conclusion. Step three: hand the raw data to three independent colleagues to recompute. If the three return three different answers, the problem lies in the definition, not the data.

The final principle of this core: every new metric must be defined before use, and it must have a denominator. A metric without a denominator is just an adjective wearing a lab coat.

Bringing the evidence chain to Vietnamese football

I have no intention of applying European parameters wholesale to Vietnamese football. But the principle travels.

The 2026 AFC Asian Cup group stage took place in January 2026 in Qatar. Vietnam lost 2-4 to Japan, 0-1 to Indonesia, and 2-3 to Iraq, leaving with zero points, four goals scored and eight conceded. Most of the subsequent debate revolved around two words: the defence.

But "weak defence" is a conclusion, not a cause. To find the cause you need to know where those eight goals came from: how many from set pieces, how many from counter-attacks after losing the ball in the middle third, how many from individual positional errors, and what the quality of the chance was in each. That data is not in the public domain in Vietnam. So the debate has to stop at adjectives, and a debate that stops at adjectives never ends.

ASEAN Cup 2026 is the opposite case, and the more interesting one to dissect.

The first leg of the final in Phu Tho on 2 January 2026: Vietnam beat Thailand 2-1, with two goals from Nguyen Xuan Son. The second leg in Bangkok on 5 January 2026: Vietnam won 3-2, 5-3 on aggregate. Nguyen Xuan Son scored seven goals across the tournament to top the scoring chart, before suffering a fracture in the second leg that forced him off.

A player scoring seven goals in a short tournament is a variable with real weight. When he leaves the pitch, everyone's model has to be rewritten.

Based on my experience watching these matches, there is a detail from the second leg that very few commentaries mentioned. After losing Nguyen Xuan Son, Vietnam no longer had an anchor up front to hold the ball under pressure. The entire attacking structure had to shift: the ball went wide earlier, the number of central combinations dropped, and the midfield had to drop deeper to receive. Winning anyway does not erase that shift. It only says the team was good enough to win inside a different structure.

That is the kind of judgment a complete dataset could confirm or refute within ten minutes. Without it, the judgment can still be right, but it cannot be verified, and a correct claim that cannot be verified is just luck retold well.

And here I have to be careful with myself.

I come from European data, and my first instinct is comparison. But before placing V.League 1's PPDA beside the Bundesliga's, I am obliged to list the intervening variables: a compressed calendar with long mid-season gaps, travel distances between rounds, temperature and humidity, pitch quality, and the way referees permit contact.

A metric built where it is 12 degrees Celsius and the grass is watered before kick-off does not carry the same meaning where it is 34 degrees and the surface is patched after the rainy season. Skip the step of naming intervening variables and I am not analysing. I am importing a prejudice and labelling it as data.

The bigger problem in Vietnam is not a shortage of advanced metrics; it is a shortage of the habit of citing sources. People argue about a match from memory, and memory is selective: it remembers the move that led to a goal and forgets the twenty that led nowhere.

Contrarian angle: the risk is not in the empty cell

The biggest risk in this trade is not an empty spreadsheet. An empty spreadsheet is honest.

The risk is a spreadsheet filled with numbers that sound plausible.

My logging system has an explicit rule: when the input is empty, the output must be empty, accompanied by a specification of what needs to be added to unlock the analysis. That rule exists for a very concrete reason. A model trained on fabricated data does not produce errors. It produces results. And results look more like truth than errors do.

At industrial scale, thousands of empty records filled with plausible football language will not break anything immediately. It simply makes everything downstream unverifiable. Readers have no way to detect it, because the prose still flows, the grammar still holds, the player names are correct, and the terminology still signals expertise to anyone who assumes terminology is expertise.

In Vietnam the local version of that risk is analysis with no source, no date, no unit and no denominator. It can still be right. But it cannot be verified.

There is a subtler second form of risk. When a new metric becomes popular, it starts being used as an adjective. "Pressing" becomes praise. "Control" becomes criticism. At some point the metric stops measuring anything; it merely colours an opinion that already existed.

I counter that with a very small habit: whenever I use an unfamiliar metric, I write its denominator in brackets immediately. If I cannot write the denominator, I have not understood the metric, and I have no right to use it to persuade anyone.

One more thing must be said about the limits of this method itself. If a reader finishes this and concludes that having numbers means having truth, I have communicated poorly. Numbers describe what happened. They do not predict the future, and they do not replace watching football. I do not predict football. I only describe probability before it happens.

And here is the counter-intuitive conclusion I want to keep: a system willing to return "not assessable" is more trustworthy than a system that always returns a score. The difference is not that the first system knows more. The difference is that the second one is hiding what it does not know.

Assumptions and lag

This piece runs on three assumptions, recorded here as procedure requires.

First: event-level data for V.League 1 and regional competitions will keep being published slowly and inconsistently in the near term. If that changes, the conclusions about the quality of public analysis must be updated.

Second: parameters built on European data need recalibration when applied to Southeast Asian football. My lag here is one season, because a comparative sample across climate and calendar conditions needs at least one full cycle.

Third: every figure cited for a specific match comes from a timestamped original record. Where the record has no timestamp, I mark the source column as "awaiting verification" and exclude it from conclusions.

Takeaway

The next round will not be decided by who owns the newest model. It will be decided by who can trace the path of each number, from raw data to final conclusion.

When the stadium falls silent, we finally hear the voice of probability. And when the data sheet falls silent, we finally hear the voice of the practitioner.

I will reopen that file. When it has numbers, I will write. Not a line sooner.

This article reflects the author's analysis based on publicly available information. Football is highly uncertain; all conclusions should be read as descriptions of probability, not as advice for any decision.