When Data Goes Silent: The Gap Inside Modern Football Analysis

WHEN DATA GOES SILENT: THE GAP INSIDE MODERN FOOTBALL ANALYSIS

One night in Incheon

On the night of 27 June 2026 I sat in front of a screen in a small apartment in Incheon. It was past one in the morning local time, and South Korea were playing Germany at the World Cup in Kazan. The match finished 2-0 to South Korea. Most viewers will remember the two late goals, Kim Young-gwon's finish after the VAR check, and Son Heung-min's breakaway into an empty net after Manuel Neuer had pushed up into the opposition half.

I remember something else.

Over the following three days I rewound the tape and counted every occasion Germany played the ball into South Korea's penalty area. The count stopped at 87. Germany's shots on target across the whole match: two. Seventy per cent of the ball belonged to them. They went out in the group stage.

I was 21 then, a third-year statistics student in Incheon. I wrote a blog post of roughly 5,000 words taking apart the way Germany's defensive line pushed high and left space behind it. Many readers found it convoluted. But an editor at a tactical analysis outlet got in touch and offered me a freelance slot. That was the beginning of thirteen years spent reading matches through data.

Six years later I sat in a meeting room in Seoul, looking at a report on screen. It had every heading. It had every numbered section. It had every cell of every table. Beneath each heading the content read the same four words: insufficient information.

The report was not wrong. It was empty. And in this industry, an empty analysis is more dangerous than a wrong one.

Context: the hunger for data and the hollow template

Professional football has entered a phase where data has become a second currency, behind broadcast rights. Every match in a top European league generates thousands of positional data points. Every K League 1 club has at least one full-time analyst, and most sides in the Asian cup-competition bracket carry three to five. A single fixture is tracked by semi-automated camera systems, by GPS vests on players' backs, by software that recognises formations.

The paradox is that the volume of data grows faster than the capacity to read it. I once watched an analysis department present a forty-two page pre-match report containing seventeen pages of tables, nine heat maps and six spatial models. The head coach leafed through as far as page five and closed it. He asked one question: where does the opponent lose the ball, and how many seconds do we have to exploit it.

Nobody in the room could answer immediately, because the answer was buried under seventeen pages of tables.

From that point I started noticing a subtler phenomenon. Reports that are full of structure and empty of substance. They appear more and more often, particularly in automated workflows. A system pulls data from a source, runs it through a template, and outputs a document with every heading in place: tactical analysis, financial analysis, risk analysis, media analysis. Every section has a table, a cell, a status line.

But when the input data is empty, the structure is still generated in full. Clubs appear in no row. No player is named. No coach is named. No match is named. The only field correctly populated is the domain label: football.

To a reader skimming the document, it looks professional. It has a contents page. It has order. It has terminology. It looks as though somebody did work.

An empty analysis does less damage than a wrong one, but it is more dangerous because it leaves no trace by which anyone can discover what went wrong.

Thirty years ago, when a coach had no information on an opponent, he knew he was blind. He sent an assistant to watch in person, phoned a colleague, or simply prepared for several scenarios. The emptiness announced itself, because it had no casing.

Today emptiness has casing. It wears the jacket of a data table. It wears the hat of a forecasting model. It signs its name in acronyms only industry insiders understand, and precisely because only insiders understand them, nobody dares to challenge it.

Mechanism: a gap does not vanish

A gap does not vanish; it simply changes its name to failure.

I learned that line not from a tactics book but from a meeting at the sports data company where I worked in 2026. I was 23. The pandemic closed K League 1 stadiums to spectators from May to August. The whole league played in silence.

I collected data from 142 matches behind closed doors and compared them with 142 pre-pandemic matches. The home win rate fell from 47 per cent to 41.5 per cent. Average goals per match rose by 0.7. Those shifts are small enough that many people dismiss them as statistical noise.

But something larger was happening that the tables did not display. An empty stand does not remove the match; it strips away the decorative layer of emotion.

In a stadium with a crowd, players receive constant signals from outside. Shouting, sighing, the sudden silence when the home side loses the ball. Those signals operate as an emotional positioning system, telling a player where he stands in the match without him needing to turn his head. When the stands are empty, that system disappears. Players have to build the map in their own heads, and most of them have never been trained to do it.

What I found in the data was not that home teams got weaker. It was that teams lost the ball more often in central areas during the first fifteen minutes of the second half. That window is normally when the crowd exerts most influence, when the home side pushes after the interval. With no crowd there is no such surge, and teams settle into waiting for each other.

I finished the report in December that year. My manager rated it highly. A Korean colleague made one observation in English: good data, published too late, no different from predicting a match after it has been played.

That sentence changed how I work. A timely analysis is worth more than a comprehensive one that arrives late. Since then I add a short section at the end of every piece setting out my data limits: sample size, time window, missing metrics.

But the bigger lesson lay elsewhere. The gap in those four months of data was not my gap. It was the league's gap. Nobody measured the silence. No metric recorded how an empty stadium altered the rhythm of players' decisions. And because it was not measured, it did not exist in any forecasting model built afterwards.

That is the first mechanism worth understanding: data does not record what it has no sensor to record. The gap is not inside the table. It is in the fact that the table has no column for it.

The second mechanism: the time between two phases

Between two phases of play, time exposes decisions the eye misses.

Based on my experience watching matches, most modern football analysis concentrates on what happens while the ball is moving. Who runs where, who passes to whom, who shoots from where. That is the easiest part to measure, because positional tracking runs continuously and the ball is a clear anchor point.

But the three seconds before possession changes is the most information-dense zone in any match, and the least analysed.

In those three seconds, four kinds of decision happen almost simultaneously. A defender decides whether to step up or drop. A midfielder decides whether to cover the passing lane or abandon it to mark a runner. A goalkeeper decides whether to hold his line or leave it. And the player on the ball decides where to look before his second touch.

Those three seconds contain no goal. No shot. No save. So they never enter the highlights package, never enter the video review, never enter the post-match report. They sit outside every standard recording frame.

I once spent two weeks manually coding the head-turn direction of four centre-backs at one K League club. There was nothing sophisticated about it: I logged every head turn, the timing, and the direction. After seven matches a pattern emerged clearly. The left-sided centre-back turned his head towards the opposing striker an average of 1.8 seconds before the ball arrived in his zone. The right-sided centre-back turned his head after 0.6 seconds.

That 1.2 second difference is not a technical problem. It is a cognitive one. The left centre-back could read situations ahead of time. The right centre-back reacted after the situation had already formed.

Across those seven matches, the team conceded seven goals from the right side and two from the left.

No metric in the standard dataset records a head turn. No camera system was designed to measure it. Yet it shaped outcomes more than every passing statistic the club published in its internal reports.

This is the second kind of gap: the gap that sits between recorded events. Football is coded into discrete events. Passes, shots, tackles, fouls. Whatever happens between those events is treated as whitespace, and whitespace does not count anywhere.

The third blind spot: surplus certainty

Germany's failure did not come from a shortage of talent, but from a surplus of certainty.

I first wrote that line in my 2026 blog post, and it remains the line I use most when discussing the psychology of big teams in World Cup group stages. Germany's squad that year was the most expensive at the tournament. They had Neuer, they had Kroos, they had everything a pre-match model needs. And that very completeness created a cognitive gap.

When a team believes it already holds every answer, it stops asking questions. They pushed their line high for 61 per cent of match time, by my measurement then, because they believed opponents could not exploit the space behind. That space was not only a gap in the formation. It was a gap in the assumption.

87 balls into the box and two shots on target. That ratio does not indicate a weak attack. It indicates an attack with no alternative once the primary plan had been neutralised. South Korea sat deep, kept the distance between their lines under ten metres, and let the opponent pass sideways in front of the box. That was not lucky defending. It was deliberate defending built on the knowledge that the opponent had no column in the plan for being stopped.

Every tactic is a hypothesis until the opponent forces you to answer. Germany arrived in 2026 with a hypothesis that had never been tested under severe conditions. When it was disproved, there was no second hypothesis, because nobody on the staff had considered a second hypothesis necessary.

Surplus certainty operates exactly like a gap. It occupies the space where questions should have been. In an analysis room, surplus certainty shows up as nobody asking where the data came from, how large the sample was, whether the sampling is biased. The report looks complete, so nobody checks whether it is complete.

Morocco 2026: the gap that was organised

Morocco did not need to control the ball; they controlled what the opponent was allowed to dream.

In 2026 I was 25, already an analyst for the outlet I had joined. When Morocco reached the World Cup semi-final in Qatar, global media devoted most of its column inches to spirit and history. I spent five days re-watching their six matches, coding every transition.

The result that surprised me was not the low possession share. It was the speed of the formation change. On losing the ball, Morocco completed their drop into a 5-4-1 structure in an average of 2.3 seconds. That number matters more than any tackling statistic, because it measures organisational capacity rather than individual effort.

Achraf Hakimi advanced an average of 58 metres per match. For a full-back that figure is extreme. It created a large space behind him. But that space was never exploited, because Azzedine Ounahi covered it under a fixed rule: every time Hakimi crossed the halfway line, Ounahi shifted across to the right flank and dropped his position by an average of four metres.

This is the point I want to stress. A gap in football does not vanish on its own. It vanishes only when somebody has been assigned to make it vanish, and that person knows exactly where to stand in every second.

Before the semi-final, Morocco had conceded only one goal in their first five matches, and that was a Nayef Aguerd own goal against Canada. No team scored against them from open play across that entire run. That record did not come from a goalkeeper in a state of grace, though Yassine Bounou played very well. It came from gaps assigned so precisely that they became part of the system rather than a defect to be concealed.

My Morocco piece ran to 3,500 words with twelve heat maps and drew 1.2 million views. On the back of it, a K League club invited me to consult part-time on tactics. But what I kept from that experience was not the view count. It was the realisation that good analysis has to answer a question about assignment, not stop at describing the shape of a formation.

VAR and the grey zone called "clear error"

During my time working in South Korea I had the chance to sit in the VAR operations room for a K League fixture. What I learned there had nothing to do with technology.

The VAR protocol rests on a standard called clear and obvious error. It sounds transparent. Sitting in the room, I realised the standard only relocates the space of judgement, it does not eliminate it. The referee on the pitch can be wrong. The VAR referee in the room can also be wrong; he is simply wrong from a different camera angle and at a different replay speed.

An incident in the penalty area, replayed at real speed, looks like a normal challenge. Replayed at half speed, it looks like a late tackle. Replayed at quarter speed, it looks like a deliberate act. The same footage, three conclusions, depending on which speed the decision-maker chooses.

And the choice of replay speed is written into no regulation at all.

This is a textbook operational gap. Everything looks procedural. There are multiple camera angles. There is a dedicated referee. There is a communications link. There is a written record. But the moment at which somebody decides to freeze the frame sits outside the procedure, and that moment decides the outcome of a match worth millions in broadcast revenue.

I am not concluding that VAR is useless. I am concluding that VAR creates a new kind of gap: the gap between the objectivity technology promises and the subjectivity people retain. Supporters see the big screen and believe fairness has been installed. What they do not see is the negotiation taking place between two referees over the headset, where one side is trying to convince the other that the error was clear.

The transfer market and organised noise

Reputation does not protect you; it only tells the opponent what to exploit.

That is true on the pitch, and it is true in the transfer market.

Over years of analytical work I have watched how deals are staged from outside. The player's agent is the largest hidden cost in the modern transfer system. Agents do not only negotiate contracts. They manufacture information. They leak it. They create phantom interested parties to drive a price. They choose the moment for a rumour to surface, and that moment usually coincides with the player's best run of form or the selling club's financial pressure.

FIFA introduced football agent regulations effective from 2026, including commission caps and disclosure requirements. In several countries the rules met legal challenge and were partly suspended by courts. The result is a market with a rulebook whose application varies between federations.

For an analyst this creates a particularly awkward kind of gap. When a club asks me to assess a potential deal, the source material I receive has usually passed through three filters: the media layer, the agent layer, and the personal relationships between sporting directors. Each filter leaves a trace, but none of them carries a label stating what it filtered.

My approach is to split the data into two clearly labelled categories. The first is verifiable match data: minutes, positions, technical metrics, physical data. The second is unverifiable market data: salary, clauses, the level of interest from other clubs. The second category I always flag as assumption, with a confidence rating attached.

Most transfer reporting on the market blends the two together. A spreadsheet places an exact minutes-played figure beside a salary inferred from media reports, and both look equally credible. That is an empty analysis wearing a data jacket.

Data only means something when we ask at the right moment; ask at the wrong moment and every figure is noise.

Goalkeeper distribution: a deified skill

Over the past decade, a goalkeeper's distribution has been elevated into a primary recruitment criterion. A keeper who passes short well is treated as the centre of a possession game. A keeper who is merely a good shot-stopper is treated as obsolete.

There is a basis for that view, but it has travelled far beyond the evidence.

When I analysed goalkeeping distribution data in one leading European league, I split deliveries into three groups by the pressure the opponent applied. No-pressure group: short passes, very high completion. Medium-pressure group: completion falls appreciably. High-pressure group: completion drops sharply and the number of long deliveries spikes.

The difference between elite goalkeepers sits mainly in the third group. But that group accounts for only about a fifth of all distributions in a season. Which means that for four fifths of the time, every keeper in that league distributes almost equally well, because nobody is pressuring them.

Yet published distribution metrics usually merge all three groups into a single figure. The consequence is that a keeper playing for a side pressed high every week will post a lower number than a keeper at a possession side, even when the underlying skill may be equal or better.

Meanwhile the basic shot-stopping metric, the thing that actually decides points in difficult matches, is rarely used as a valuation criterion.

I am not denying the role of distribution in modern football. I am questioning how much of it the market is paying for. A keeper with a high distribution figure commands a high price, and sometimes that price reflects a club buying a function from his previous system rather than a capability that transfers into a new one.

The execution blind spot: a full frame with an empty core

Back to the report in Seoul.

I have spent most of this piece discussing tactical gaps. But there is another kind of gap that sits above all of them, and it belongs to my profession rather than to any club.

It is the gap inside the analysis process itself.

A report with every heading, every numbered section, every formatted frame, but no underlying data, cannot be detected by skimming. It is detected by asking one question: where is the source data for this section.

And in most meeting rooms that question is never asked, because the presenter appears to have done work, and the audience does not wish to appear distrustful.

I have seen a pre-match report presented over forty minutes with seventeen pages of tables, after which an assistant coach discovered that the entire opposition analysis had been copied from the reverse fixture of the previous season, in which the opponent had fielded a different line-up under a different coach. The report was not wrong because it invented data. It was wrong because nobody checked whether the data was still true.

That is the nature of this class of error. It is not a fabrication error. It is an inheritance error. A frame was created, a template saved, a process established, and afterwards everyone reused it because it looked professional.

The only defence I have found after thirteen years is to state the limits inside the document, in a position that cannot be skipped. Every report I write carries a section setting out sample size, time window, data source, missing metrics, and which conclusions are hypotheses rather than findings. That section does not make the report look less professional. It makes the report verifiable.

And when a document is not verifiable, whether it is right or wrong becomes a matter of luck.

A conclusion you can test in the next match

In the next match you watch, try one small exercise. Pick one team, note the moment they first lose the ball in a central area, and note the position of both centre-backs three seconds before possession changes.

Do that across three consecutive matches. If a pattern repeats, you hold something no data table hands you ready-made. If nothing appears, you hold a gap correctly named, rather than a gap filled in with guesswork.

A gap does not vanish; it simply changes its name to failure. And data only means something when we ask at the right moment.

Data limits for this article: personal figures were recorded during live observation and manual coding, the sample is small, and it is not statistically representative of any full league. The figures concerning Morocco at the 2026 World Cup and the 2026 K League season were cross-checked against public data, and errors may persist at the level of individual phases of play.