The Report Is Never Written in Advance: Reading Fouls and Football's Habit of Early Conclusions

Opening

September 2026, Groupama Stadium, first half of Lyon – Marseille. I was sitting in the section reserved for league discipline reporters, my notebook open on a page ruled into five columns: minute, player, type of foul, referee's decision, video note. Dimitri Payet received the ball on the left channel, was checked shoulder-to-shoulder, went down, got up, carried on. The referee played the advantage. I wrote one line. Four minutes later Payet went in from behind, no contact with the ball, I wrote line two. Line three came in the 39th minute, after a shirt pull in midfield that the assistant had already flagged.

That night I closed my report on the figure three.

The next morning, the Ligue 1 disciplinary department returned it with one short line: Payet's foul count did not match the video. Four, not three. The foul I missed was in the 31st minute — Payet blocked a pass with his hip, the referee did not blow, the ball changed direction, play rolled on. No whistle. No card. Not a single shout from the stands. A passage so unremarkable that I discarded it from memory before I had even written it down.

For the next four weeks I rewatched the full footage of twelve Marseille matches, frame by frame, and built a personal error code table of forty-seven entries. Not because a wrong number had got me reprimanded. Because I had seen the mechanism of the mistake: the data written down is not the data that happened, it is the data the writer still remembers.

That error in the 2026 World Cup qualifiers taught me this: the report is never written in advance.

Context: how many layers make up a disciplinary file

In Ligue 1, every matchday feeds the disciplinary committee three parallel streams. The first comes from the referee and the two assistants: the match report, with an error code for each incident. The second comes from the VAR room: time of intervention, type of incident, whether the decision was overturned or upheld. The third comes from discipline reporters like me: independent observation, no authority, only an obligation to cross-check.

These three streams do not align on their own. They align only when someone sits between them and verifies every line.

My error table has forty-seven codes in four groups: off-ball contact, counter-attack prevention, disputes with the referee, and behavioural offences. Each code carries a minimum definition so that someone else can re-check it. Code 14 — blocking with the body after the ball has left the opponent's foot — is never recorded unless I can establish both the moment the ball left the foot and the direction of the blocker's movement.

The rule I set myself after those four weeks of Marseille footage: a number only enters the report when at least two independent sources confirm it — my eyes in the stand and footage from a different camera angle. If only my eyes, it is an observation. If only the footage, it is evidence, but not yet a judgement.

There is an uncomfortable point about this trade that few people state. The pressure does not come from missing data. The pressure comes from the fact that the report must still be filed on time, whether the data is complete or empty. And when a template is designed to always return a conclusion, it will always return a conclusion — even when there is nothing inside to conclude from.

The five layers of a match file

Over the years I have systematised every match file into five layers. The identity layer: competition, round, absolute date, referee, assistants, VAR. The entity layer: every team, player and coach named. The fact layer: three to ten verifiable statements, each tied to a source and its own credibility grade. The quantitative layer: foul counts, cards, VAR interventions, and where available xG, xGA, PPDA. The time layer: whether this is a same-day event, a season-long trend, or a multi-year structural story.

These five layers are not an administrative checklist. They are a load-bearing structure. Without the identity layer a file cannot be cross-referenced; without the entity layer it cannot assign responsibility; without the time layer every conclusion has an undefined expiry date.

The time layer is the most neglected. A VAR overturn rate measured over one matchday may be a single-day anomaly, or it may be the peak of a half-season trend. Filing both into the same cell is a methodological error, not a numerical one.

When a layer is empty, conclusions are still produced

I once received a file from a young colleague: proper headings, full tables, exactly three conclusions per section as required. But the entity layer was empty. No team named. No player named. No competition named.

The striking thing was not that the file was empty. The striking thing was that it still contained three conclusions per section, and those conclusions read very smoothly.

This is the most dangerous mechanism in analytical writing. When the template demands three conclusions for a section, the writer will find three conclusions. If the data cannot supply them, the writer supplies them instead. Nobody deliberately fabricates. But a mould that always demands output forces the writer to fill the input.

In football, the symptoms are easy to spot. A disciplinary file that names no match cannot be appealed. An analysis with no absolute date cannot be contradicted. A piece of commentary that names no player cannot be checked against footage. These three document types share one property: they are immune to verification, and that is precisely why they exist.

The only correct conclusion when the input is empty is to state clearly that the input is empty. I call that the discipline of the blank cell.

A fine mesh, and the lesson of the 2026 World Cup

My method for covering the 2026 World Cup was a net: small mesh, no fish missed.

In June 2026, the desk assigned me a feature on yellow-card sanctions. The easy route was to follow the big matches — France against Argentina, the duels featuring Mbappé and Messi, games the audience had already watched and merely needed their feelings confirmed on. I took the hard route: the fourteen lowest-scoring group-stage matches. Games to which the international press barely bothered sending anyone.

The result lay where nobody expected. Iran under Carlos Queiroz had the highest rate of counter-attack-preventing fouls in the tournament: twenty-three in three group matches. Not destructive fouling. Calculated fouling, committed at the exact moment a counter-attack had not yet formed, in the exact area of the pitch where a referee has the worst sightline.

The European Referees' Council cited that piece. So Foot invited me to contribute. But the real reward was not those two lines. The real reward was the proof of something I had long suspected: the most valuable data in football usually sits in the matches nobody can be bothered to watch.

Those fourteen games were not glamorous. No superstars. No major controversies. That is exactly what made them a clean environment for measuring tactical behaviour, unpolluted by crowd emotion and scoreline pressure.

A net does not need to be wide. Its mesh needs to be small.

Useless running also produces beautiful numbers

There is a paradox in fitness data that I have encountered repeatedly, at athletics championships and Olympic Games I have commentated on. Distance covered and sprint counts are packaged as effort indicators. A midfielder who covers twelve kilometres in a match gets lifted onto an effort leaderboard. But useless running also produces pretty numbers. Chasing a ball that has already gone, running towards a teammate who is already marked, running to fill a space nobody is attacking — all of it adds to the total.

Referee data suffers from the same disease. The number of fouls called does not measure accuracy. A referee who calls many fouls is not a good referee; he may be a referee who has lost control of the match. A referee who runs a lot is not well positioned; he may be reading the direction of play badly. The correct criterion for a referee is not how often he blows, but the incidents he should have blown and did not — and the ones he should not have blown and did.

The 31st-minute incident at Groupama belongs to the first category. It appears in no statistical table. It appears only when someone sits down and rewinds footage from another camera angle.

VAR and an unresolved time layer

VAR adds a new event stream: when a review began, how long it lasted, the incident type, the outcome. Technically, this is a far better data source than what came before. No more waiting for someone to rewind tape; the data generates itself with every passage of play.

But self-generating data is not self-verifying. The number of VAR interventions only means something once placed inside the time layer. Nine interventions in one matchday is an anomaly requiring investigation. Nine interventions per matchday sustained across half a season is a characteristic of the league, and it must be handled entirely differently. Without the time layer, the number stands alone and becomes a tool for whoever wants to use it.

From the stands, VAR does not eliminate controversy. It relocates it. Where people once argued about the referee's eye, they now argue about the line and about which frame was chosen to draw it. The object of dispute changed; the volume of dispute did not.

That is why I never treat VAR as a revolution in fairness. It is a revolution in traceability. The two are different things, and conflating them is the source of most bad refereeing commentary in a normal season.

Which dimension carries weight is itself a finding

For years I made the same methodological error: applying one template to every kind of story.

A transfer story heavily activates the dimensions of club finance, registration rules, contract terminology and media cycles. The tactical dimension in that kind of story is nearly empty. A match review inverts the weighting entirely.

Whether a given dimension carries weight is itself a finding. If I force a piece of referee reading to deliver three conclusions about transfers, I am forcing the data to say something it does not contain. And when forced, the data does not stay silent. It speaks wrongly.

In my work, the difference between a usable disciplinary report and one that gets returned rarely lies in the volume of data. It lies in whether the writer accepts letting the data determine the weighting.

The counter-intuitive angle: another layer does not automatically add another truth

Modern football tends to settle disputes by adding process. Add VAR. Add a fourth official. Add a post-match explanation file. Add an error code table. The sense of safety comes from the number of layers.

But a pipeline designed to always produce an output will always produce an output, regardless of the input. In the worst case, process does not control the quality of conclusions. Process only controls whether conclusions are generated on schedule.

Running alongside that is a larger problem traditional football underestimates. Esports betting is eroding competitive integrity far faster than traditional sport, simply because the rulebook is running behind reality. In that environment, a floating number with no identity layer, no source and no absolute date is a favourite meal for the market.

People worry about referees being bought. Fewer people worry about referee data being misread systematically. The second risk occurs weekly, requires nobody to approach anybody — only a report template that demands conclusions before the data has arrived.

Takeaway

I propose a minimum verification gate, placed before any match file becomes a report: at least one identifier, at least one named entity, at least one verifiable incident with a source, and at least one absolute timestamp. Fail that, and the file stops. A blank cell is recorded as a blank cell, and that is a valid outcome.

A net with one broken mesh lets the fish through that exact mesh.

My method for covering the 2026 World Cup was a net: small mesh, no fish missed.

The question that remains belongs to those currently holding the pen: if every conclusion had to wait for the final whistle, would we still have the patience to write — or have we grown used to filing the report before the match ends?