The Empty Spreadsheet and Vietnamese Football's Habit of Filling It With Stories

Minute 44, Nha Trang stadium. I was sitting in the third row of Stand B, my eyes fixed on the opposing right-back's feet rather than the ball. Three times in seven minutes, he dropped back half a step before the ball arrived. Three times, the right-sided centre-back turned toward the near post instead of covering. The distance between the two shirts stretched from four metres to seven, then eight. Nobody pointed. Nobody shouted. The first half ended 0-1, and in the stands people talked about an individual error.

At half-time I went down to the technical room and rewound the tape twice. Fourteen attacking sequences from the opposition in the first forty-five minutes. Fourteen times the ball travelled into exactly one gap. I redrew the diagram on a sheet of A4, circled that zone in red pen, and proposed switching from 4-4-2 to 3-5-2. In the second half we won 3-1. Dangerous sequences entering the red circle dropped to two. I did not celebrate. I added one more defensive variant for the following round.

Years later, what brings me back to that sheet of paper is not the red circle. It is an empty column sitting right beside it. My coding sheet that day had a field called "passes into the red zone". That column stayed blank for the entire first half. Not because there were no passes — I had just counted fourteen sequences. The column was blank because I had never managed to define what "a pass into the red zone" actually meant. An empty cell in a spreadsheet has never had a single meaning. That is where the story starts.

Context: a league full of cameras, empty of data

In 2026, V.League stopped because of the pandemic. No spectators, the stands as silent as a blank sheet. I made no long-term plans. I sat down and coded all 1,247 corner situations of the 2026 V.League season. The result: one goal for every thirty-seven corners. The South-East Asian regional average at that time sat at roughly one in twenty-five. The league's conversion rate was nearly a third below the regional baseline.

I sent the report to a club in Nha Trang without anyone assigning me the work. What came back was not a rejection but a question: "So what should we do?". That is the best question a report can receive.

But to answer it, I had to deal with something bigger: how much trustworthy data does V.League actually have?

Cameras are plentiful. Every round produces hundreds of hours of footage, from the main camera to the behind-goal rigs. Pay television, digital platforms, social media — everything is replayed. A single V.League goal can be watched from seven different angles within two hours.

Footage is not data. Footage is raw material. Data is something that has been named, defined, counted, and made verifiable. Between those two things lies a gap that most Vietnamese clubs have never measured.

I once asked a V.League coach what metric he used to assess a central midfielder after a match. He answered quickly: "You just look and you know." I asked what exactly you know. He went quiet. Not because he was unskilled. He had twenty years of professional eyes, and those eyes genuinely see things a spreadsheet does not. But when pressed to the level of specificity, to a unit of measurement, the professional eye runs out of words.

This is nobody's personal failing. It is the failing of a system that was never designed to answer the question "why".

The gap does not create itself

Back to the red circle in Nha Trang. People called it "the right-back's error". That phrasing is wrong because it turns a system problem into a personal moral problem.

Look at the mechanism. When the ball is on the opposite wing, the defending team must slide its block across. The right-back drops toward the near post to protect the space in front of goal. The right-sided centre-back has two options: cover wide, or hold position to protect the central corridor. If the defensive midfielder does not drop into the inside channel in time, the centre-back is forced to hold the middle. The result: a gap opens between the two of them, four to eight metres wide.

That gap is not caused by a lazy player. It is the product of three correct decisions placed next to each other inside a system with no spare cover.

That is why I do not trust reports that list individual errors. A pass that lands two metres off is not a technical error; it is a crack in the whole perceptual system. People see the misplaced pass because it shows up on the screen. They do not see the defensive midfielder who was two seconds late dropping back from the thirtieth minute, because nobody counted that.

Fourteen sequences in forty-five minutes, all funnelled into a single gap — that is a pattern, not a run of accidents. Had only two of those fourteen gone into the gap, I would have called it random. Fourteen out of fourteen leaves no room for randomness. Repetition is the fingerprint of a system, not of an individual.

In the second half we switched to 3-5-2. Geometrically, that did three things at once: it added a centre-back to fill the inside channel, it pushed the two wide players deeper to seal the flanks, and it kept three midfielders centrally to block the line-breaking pass. Dangerous sequences entering the red circle fell from fourteen to two. Not because the players ran more. They ran less.

But here is the part I kept to myself for years: I do not know whether the remaining two sequences were due to the new system working, or due to the opposition changing their attack in the sixtieth minute. I had no data to separate those two possibilities. Once again, an empty cell.

Two meanings of an empty cell

This is the thing I most want Vietnamese football readers to understand clearly, because I see it misread constantly in match reports.

An empty cell in a dataset has two completely different meanings. The first: the event did not happen. The second: the event did happen, but nobody recorded it, or recorded it under the wrong definition.

These two meanings lead to opposite conclusions. If you read it wrongly, you will conclude the team did nothing at all — when in fact the team did plenty, and you simply lacked the instrument to see it.

In the Nha Trang case, the "passes into the red zone" column was blank for the second reason. The event was real, fourteen times over. The recorder had no definition.

I have seen the same thing at national-team level. After major matches, the flood of analysis online tends to use phrases like "controlled the ball better" without saying in which zone, over which period, under pressure from how many players. Total possession share is one of the most useless metrics in modern football. A team holding sixty percent of the ball but holding most of it in its own half is being pushed back, not controlling.

I often ask myself why this misreading is so durable. The answer lies elsewhere: it is convenient. A single aggregate number is easier to write than a zone map. And in the writing trade, whatever is easy wins.

Corner data: the only thing that never takes a half-time break

When the season stood still, the corners kept rolling through my spreadsheet. In 2026, with the whole league paused, I had the time to do what a normal season never allows: to code every corner of an entire season.

The striking result was not the conversion number. I split the data along two variables: delivery position and marking scheme.

On delivery position, most V.League teams deliver to the near post. The near post has the advantage of a short flight path, fast arrival, and a defender who struggles to turn. But the near post is only effective when at least two attackers move diagonally to drag the defensive line out of the landing zone. In my data, many teams delivered near-post with only one body committed. The result: the ball arrived, but the receiver was already marked.

On marking schemes, I found a systemic hole. V.League teams defending corners typically used pure man-marking, or man-marking plus one player on the post. Very few teams positioned anyone to block the second-ball zone — the area roughly twelve to eighteen metres from goal. The consequence is that when the ball breaks loose, the attacking side usually controls it first, because the defending side has pulled everyone inside the five-metre ring.

The second ball is where most V.League corner goals are actually born. Yet match reports still count "corners: 7-3" and call that statistics.

Corner numbers do not lie, but they stay silent until you ask the right way. Ask "how many" and they go quiet. Ask "where did the ball go, who moved, who stayed" and they open up.

I once ran a small experiment with a group of young coaches. I showed them five corners from the same team and asked them to predict which ones produced a chance. Nobody got more than two out of five right. Then I showed them again, with a requirement to count the number of defenders inside the twelve-metre zone at the moment the ball left the deliverer's foot. With that number in hand, accuracy rose to four out of five.

The difference between the two attempts was not the eye. It was that in the second attempt they had a specific question.

The problem is not a lack of data, it is a lack of questions

For years I have heard the same complaint: Vietnamese football lacks data. That is true, but it is true in a useless way, because it names a deficit without naming a remedy.

Data does not emerge from a camera lens. Data emerges from a question. You cannot code "passes into the red zone" if you have not defined the red zone. You cannot count "a defensive midfielder two seconds late" if you have not decided that the timing of the drop is a meaningful variable.

Which means the correct order is question first, data second. In Vietnam, that order is usually reversed. People buy the software first, then scramble to work out what questions the software can answer.

Based on my experience watching V.League matches across many seasons, I see a recurring pattern at clubs that are just starting to do analysis. They build very pretty dashboards. They have passing data, duel data, running data. But when the coach asks "so how do we play next match", the dashboard cannot answer. Because a dashboard describes what happened; it does not describe what should happen.

The distance between description and recommendation is the distance between a secretary and a member of the coaching staff. Both are necessary. Only one of them sits in the tactical meeting.

The wing problem: when a whole league makes the same mistake

There is one subject I have pursued longer than any other, and it connects directly to squad structure in V.League: the fate of the traditional winger.

Recent years have pushed wingers inside. This is a global phenomenon, spreading from the leading football nations down into regional leagues. The justification is sound: an inverted winger creates a numerical advantage in the centre, opens up shooting on the stronger foot, and drags the opposing full-back out of position.

But one variable gets ignored when this model is imported into V.League: width.

In top leagues, when a winger inverts, the full-back advances to hold the width. They can do that because they have the physical base to run all match at high intensity and they have a defensive midfielder covering behind them. In V.League, many teams adopt the inverted model without either condition. The result is a system with nobody holding the touchline. The attack piles into the centre, where the density of players is already high, and loses its ability to stretch the opposing defensive line.

I do not believe in resurgence, I believe in placing the ball back where resurgence becomes possible. For Vietnamese football, that place is the flank. It is where the fewest players are, where the pressure is lowest, and where the least complex skill is required to manufacture a valuable situation.

Erasing the traditional winger in V.League is an environmental error, not a tactical one. A model that is right where talent density is high can become wrong where talent density is low. Optimise for the environment, not for the theory.

The consequences are measurable. At V.League teams without a genuine touchline player, the volume of passes delivered from wide channels into the penalty area drops noticeably, while lateral passes in the centre rise. The lateral central pass is the least valuable pass in football: it advances little toward goal and is easy to intercept.

In other words, many Vietnamese teams are shrinking their own pitch.

The inverted winger: the hidden cost

I want to be clearer about the cost of the inverted model, because analysis usually only sees the benefit.

When a winger inverts, he drags an opposing full-back with him. That is the benefit. But he also drags himself into a zone that is already crowded — that is the cost. In the centre, more players means less space per player. A central midfielder needs roughly two square metres to handle the ball under pressure. If the inverting winger consumes that space, the central midfielder loses the ability to turn.

In top leagues, this cost is offset by speed of ball handling. Players there decide in about a third of a second. In V.League, that window is longer. Which means the same model carries a higher cost in V.League.

There is a simple test anyone can run. Pick a V.League match and count how many times the team holds the ball in a wide channel within the ten seconds before losing it. Then count how many times they lose the ball centrally. If the central loss rate is higher than the wide rate, that team is being suffocated by its own model.

I have run this count on a handful of matches and the results caught my attention. Many teams lose the ball centrally at a markedly higher frequency, yet keep the inverted structure for the whole match. Asked why, the answer is usually: "This is how we train".

That is the answer of habit, not of analysis.

Metric transfer: when someone else's tool measures the wrong thing

There is a deeper layer than the winger story and the corner story: metric transfer.

Modern football analytics metrics are built on data from Europe's leading leagues. Expected-goals models are trained on hundreds of thousands of shots from those competitions. The metric for passes allowed per defensive action is defined around the tempo of matches played there.

That tempo differs from V.League. Not slightly — structurally.

In leading leagues, a team's pass volume per match is usually high. In V.League, that figure is lower, turnover counts are higher, and transition situations are more frequent. A metric built on pass volume will automatically produce wrong results when applied to a match with a different tempo.

A concrete example: when you evaluate a defensive midfielder using passes allowed per defensive action, you assume the opposition passes a lot. If the opposition plays direct and passes little, that metric will be very low — but it does not tell you the midfielder played well. It tells you the opposition did not pass.

People shine the light on winners, I shine it on where they stumbled. With metrics I do the same: I look at where the metric fails, not at the moment it produces a good-looking number.

The right response is not to reject imported metrics. It is to recalibrate them with domestic data. A V.League expected-goals model needs training on V.League shots. A definition of pressure needs building on the number of players V.League actually commits to the ball zone.

This recalibration requires something Vietnamese football does not yet have: open event data, long enough and clean enough. Without it, all analysis is translation.

The management layer: nobody wants a number that talks back

At this point I have to say something rarely said in Vietnamese football analysis.

The biggest problem with data analysis in V.League is not a lack of tools. It sits at the decision-making layer.

A verifiable number creates accountability. If you have data showing a player running two seconds late in the inside channel across fourteen consecutive sequences, then continuing to start him becomes a decision that can be challenged. If you have no number at all, that decision can still be explained away by "experience" and "belief".

This is not a Vietnamese speciality. It is true in every football nation. But it is more visible where club structures have not yet separated the decision-maker from the person accountable.

I have seen a good report shelved. Not because it was wrong. Because it was right in an inconvenient way.

My method in that environment is simple: always offer two options. One grounded in data, one grounded in the existing structure. Judge neither as better. Just show the cost of each. When people see the cost, their decision changes. Not because they were persuaded morally, but because they do not want to pay.

Optimise for the environment, not for absolute value

Here I want to reconstruct a reading habit that is common and harmful.

When I hear someone praise a player, I usually ask back: in which environment. The same player, with the same skill set, can be optimal in one system and anti-optimal in another. This is not new in theory, but in practice it is forgotten constantly.

A central midfielder who is good at holding the ball and turning will shine in a possession system with plenty of short passing options. The same player, in a direct system, will be repeatedly exposed because his handling time is too long for the required tempo.

This is why clubs should assess players by a "fit index" rather than a "total index". A fit index measures the match between a player's characteristics and the system's demands. A total index only measures average quality.

I know building a fit index is much harder, because it requires you to define your system first. Many Vietnamese clubs have no document at all describing their system. They know they play 4-2-3-1, but they do not know how they intend to score.

The transfer market: the youth bubble and the price of a belief

There is a subject I track that connects directly to the data layer.

Over the past few years, the price of young Vietnamese players has risen at a rate that does not correspond to the top-flight minutes they have accumulated. Large contracts for players who have not yet played even fifty professional matches are becoming common. This is a naked gamble, not a strategy.

But seen from the data layer, it is understandable. When there is no long enough database for valuation, people value with stories. The player with the better story costs more, not the player with the better data.

Which means: the data deficit in V.League does not merely weaken analysis. It directly distorts prices in the transfer market.

An open event dataset, long enough and clean enough, would cool emotional valuation. When you can show that a twenty-year-old has played twelve hundred minutes in the top division with stable output across two seasons, his price anchors to a reference point. When nobody counts those minutes, the price anchors to a memory.

This benefits both buyers and sellers. The problem is it does not benefit intermediaries, who live off information asymmetry.

A map of unwritten gaps

I have one professional habit: before every match I draw a map of my own team's gaps. Not the opponent's gaps — everyone does that. My own, mapped before the opponent exploits them.

That map lists three things: which spaces of ours are not controlled by a specific player, which periods of the match we are most likely to lose structure, and which set-piece situations we have no contingency for.

What I have found after many seasons working in V.League is this: most goals conceded do not come from unknown gaps. They come from known gaps that were never recorded, because the person who knew kept it in his head.

In a tactical meeting, knowledge held in three heads can be preserved. Knowledge held in one head depends on that person being present. And in football, that person can be sent off, injured, or sold.

This is why I always translate every observation into a transferable format: diagram, distance, timestamp, rate. In the dressing room, I do not listen to voices, I read the positions of the shoes. With the gap map, I read the positions of the spaces nobody has claimed.

The biggest blind spot: believing data will save you by itself

Here I want to reconstruct a counter-argument I have had to face myself.

Since data analysis became a trend, a new belief has spread: with data comes a conclusion, with a conclusion comes a result. This belief produces a consequence few anticipate — it creates a new layer of storytelling, dressed in numbers.

That is the blind spot. A dense dataset can be more persuasive than an empty one, but it is not automatically more correct. If the variables were chosen on a false assumption, the output is simply a mistake presented neatly.

I have fallen into this trap. After coding a large dataset, I found a very elegant pattern and wrote it into a report. A week later I discovered that pattern appeared because my data was missing a third of the matches of one team. The elegant number was not wrong arithmetically. It was wrong definitionally.

Which means there are two kinds of error in football analysis, and the second is more dangerous. The first is being wrong when you have no data. The second is being wrong with full data, because you asked the wrong question from the start.

With the first kind, people can spot it, because the emptiness is visible. With the second, nobody spots it, because everything looks sealed.

If you see nothing at minute 60, rewind from minute 59

There is another habit I want to share, because it belongs to the methodological layer, not the tactical one.

When analysing a match, I never start at minute one. I start from the final result and trace backwards to the cause. The goal at minute seventy-three is the end point of a causal chain. I rewind from seventy-three to seventy-two, then step gradually back. At each step I ask myself: if this person had decided differently by a fraction of a second, would the chain still lead to the goal.

This backward reading is unnatural. The human eye is trained to read forward, along the timeline. Reading backwards requires fighting your own habit. But it is necessary, because a match is not structured along a timeline. It is structured along causal relations.

In the France-Croatia World Cup final in Russia in 2026, I tracked and coded all sixty-four matches with my own statistical sheet. In the final I found that Luka Modric's high-intensity running figure fell by about twelve percent after the sixtieth minute. At the same time, France's attack repeatedly switched direction into the zone Modric had to cover. Croatia shifted into a deep defensive block, but the dropping midfielders could not maintain their spacing, opening up large gaps in the centre.

A twelve percent drop in intensity does not sound like much. But it was enough to change decisions. A midfielder who no longer has the energy to reach the receiving point will choose a safe position, and his safe position is someone else's gap.

I tell this example not to compare Vietnamese football with world football. I tell it to make one point: if you see nothing at minute sixty, rewind from minute fifty-nine. Many V.League problems are seeded long before they appear on the scoreboard. A halo does not go out overnight; it begins to crack at a specific minute nobody marked.

Why Vietnamese analysis is hard to verify

There is a structural problem in how football analysis is written in Vietnam that I think needs naming.

Most post-match analysis is written with words that cannot be verified. "Spirit", "character", "class", "composure" — these words have high expressive power, but nobody can define them on a scale.

When a concept cannot be verified, it cannot be refuted. And when a concept cannot be refuted, it survives in the discourse forever, even when it is wrong.

This is why I try to translate every observation into a verifiable format. Instead of "the defence looked rattled", I write "the distance between the two centre-backs grew from four to eight metres over ten minutes". Instead of "the attack was stuck", I write "the team lost eighteen of its last twenty-four passes in the centre".

This style is drier. It does not read well. But it has one property that elegant analysis lacks: next week people can check whether it was right or wrong.

And in analytical work, the capacity to be proven wrong is the only thing that keeps you honest.

People shine the light on winners

There is a pattern I see repeated in how Vietnamese media handles football: the light shines on the winner.

A team that wins three in a row receives a wave of admiring analysis. A team that loses three in a row receives a wave of critical analysis. But both kinds of article usually describe the same thing: the result. They are written after the result exists, and their content is selected based on the result.

This is the statistical problem known as selection on the dependent variable. When you choose your sample based on the outcome, you cannot draw causal conclusions from it.

People shine the light on winners, I shine it on where they stumbled. Not because I want to bring anyone down. Because the stumble stays in the same place through wins and losses, while the light changes direction.

The only way to see a structure is to observe it in both states.

The institutional layer: data is infrastructure, not a service

A question I have asked myself for years: who should be responsible for building event data for V.League?

At club level, it is not economically viable. A single club coding an entire season would spend resources it does not have, and the data produced would only serve itself.

At league level, it creates shared value. A complete event dataset for a season is an asset every club can use, and shared use does not reduce its value — the more users, the more the data standard is stress-tested.

But infrastructure data does not produce results within one season. It is an investment that pays back over years. That makes it permanently ranked behind investments with immediate returns.

This is the class of problem markets do not solve. It needs a decision at league governance level, and it needs to be sustained long enough for the data standard to stabilise.

I hold no illusions about the speed of this. But I know the cost of not doing it: every season that passes, a season of data is lost. And football data is a non-renewable resource.

You cannot go back to the 2026 pitch to code it again.

An afternoon in Nha Trang, and another empty cell

I want to close with a small on-site observation.

A few months ago I was sitting in a coffee shop near the beach in Nha Trang, rewatching an old V.League match on my laptop. Next to me, two middle-aged men were watching the same match on a phone. They argued heatedly about one passage of play: should the winger invert or hold the touchline.

The first said the winger must invert to create numbers in the centre. The second said the winger must hold the touchline to stretch the defence. Both were right under a certain condition. And neither had the data to prove what that condition was.

The argument lasted fifteen minutes, then ended with a sentence I have heard many times in my life: "Forget it, talking about this gets us nowhere".

I sat quietly. Not because I agreed. Because I knew both men were right, and what made the argument unending was not the opinions. What was missing was a dataset.

Between two correct opinions without data, people choose by sentiment. In football, sentiment usually follows the story that already existed.

Minute 60 is not the end point, it is the start point

I do not know how many years V.League needs to build a usable data infrastructure.

I know something else. Every correct analysis begins with a specific question, and every specific question begins with an observation that was recorded.

Which means the first task is not buying tools, not hiring specialists, not writing reports. The first task is choosing one thing to count, and counting it properly across a whole season.

If you choose well, you end up with a small but clean dataset. A small clean dataset is worth more than a large empty one. Because a clean dataset can be expanded. An empty dataset can only be filled with stories.

Next match, I will sit in the third row of Stand B in Nha Trang and do the old job again. I will not count goals. I will count the distance between my team's right-back and right-sided centre-back at the sixtieth minute.

If that distance is seven metres, I know I will have one more cell in the spreadsheet. And this time, I have defined it.