The Empty Dossier: Discipline in Tactical Analysis

At 7:40 in the morning in Lyon, the tram along rue de Marseille slips through the gap in the office window. I open the PDF that has just arrived from the data desk: nine pages, nine analytical dimensions, and every substantive field carrying the same sentence — insufficient information, cannot assess. No competition named. No club named. Not a single minute of football recorded. The cover page is blank on all four lines: title, source, article type, author stance.

What matters is that before I finished the second page, three backup headlines had already formed in my head. One about a midfield crisis in Ligue 1. One about the price of selling academy players. One about a manager under pressure after four winless matches. All three sounded plausible, all three were timely, and none of them rested on anything beyond my own imagination.

This happens more often than people assume. A corrupted data file, a truncated match log, a source page returning an empty shell because of a login wall. For a content professional, that is the most dangerous moment of the day — the system hands you a blank space, and the job demands a finished piece.

The industry does not reward silence

I entered the trade in 2026, starting at a local radio station where every bulletin ran two minutes and the editor counted seconds. Four years hosting a football night show taught me something no classroom did: broadcast does not wait for verification. You go on air at nine, you need content at nine.

That tempo transferred intact to digital publishing, only faster. A match ends at 22:50, and by 23:30 there are twenty versions of it online. Within that rhythm, four article archetypes dominate the market: match and tactical analysis, transfers and contracts, managerial change, and finance or compliance. Each carries its own pressure, and each can be filled with speculation if the writer does not stop himself.

I once thought caution was a personality trait. I later understood it is a process — one that can be written down, audited, and most importantly, refused.

Where a blank space comes from

When an extraction pipeline returns an empty payload, the cause almost always sits at the retrieval stage, not the reasoning stage. The source page returns a JavaScript shell, a cookie wall, or an error code; the parser receives nothing; and the model downstream receives an empty input.

The diagnostic signature is clear: even the cheapest fields are blank. Title blank, source blank, publication date blank. A functioning system nearly always captures a title, even behind a paywall. When the title itself is missing, the problem is not that the article is hard to analyse — the problem is that the article never reached the analyst.

The second signature is subtler: template instructions surviving inside content fields. The entity field still reads "identify from the information points above" instead of naming a club. That is the fingerprint of a form that was never filled, pushed outwards wearing the costume of a finished report.

A report that looks complete while containing only empty template fields is more dangerous than an obviously blank one, because nobody thinks to check it.

In football the same mechanism repeats at a smaller scale. Automated post-match tables sometimes render a zero where data is absent. A holding midfielder who made no tackles because his side held 68 percent of the ball and the opponent never advanced is recorded with zero tackles, and the reader concludes he played badly. That zero is a technical lie, not a tactical fact.

Saying "insufficient data" is a professional skill

Applied statistics has a convention called missing-value handling. The principle is simple: when there is no data, record that there is no data, and never interpolate a plausible number. Interpolation is only valid with a known distribution, and in football that almost never exists.

Tactical analysts across Europe apply this convention constantly without naming it. A player returning from an ACL injury with 240 minutes in four months produces noisy pressing metrics. A new manager who changes shape after three rounds offers a sample too small to support any claim. A league that alters stoppage-time rules breaks year-on-year comparison entirely.

The minimum inputs for a football analysis look modest. Tier one: title, source, date, and at least three discrete factual assertions. Tier two: named clubs, players, coaches, competitions, and every relevant financial figure. Tier three: a source-reliability grade and a time-sensitivity assessment. Without tier one, everything downstream is meaningless.

Three data layers and their blind spots

The three most-quoted metrics in the game all carry structural blind spots.

Expected goals measures chance quality by position and angle, separating process from outcome. But xG does not know who is shooting. A twelve-metre attempt by an elite striker and by a defender who wandered into the box are scored almost identically. Its defensive twin, xGA, cannot distinguish a back line that was torn apart from one that deliberately invited the opponent into harmless areas.

PPDA — passes allowed per defensive action — proxies pressing intensity, with lower values meaning more aggressive pressure. But PPDA averages the whole match, so it cannot answer the question that matters most: how many minutes did this side actually press, and at which minute did the structure collapse?

Heat maps deserve the most suspicion. They have become the new fortune-telling of the industry: a beautiful image, easy to share, and almost always concealing a player's real role. A central midfielder glowing red across the middle third might be the tempo regulator, or he might be the man repeatedly dragged out of position, exposing the space behind him. Same image, opposite meaning.

I trust the pressing map more than the post-match quote. A pressing map records direction, trigger timing and who gave the signal — things that cannot be staged in a press room.

The diamond in the rough and the discipline of rewatching

In September 2026, aged eighteen, I attended Lyon Duchère against Jura Sud in the French fourth tier at Stade Balmont. Mohamed Sarr, a twenty-year-old central midfielder, recorded 58 touches, 51 successful passes from 55 attempts at 92.7 percent, six interceptions, no goals and no assists. The next day's reports mentioned only the striker who scored twice.

I spent two weeks rewatching Duchère's previous four matches, counting passes by hand, and found that 80 percent of their dangerous progressions passed through Sarr's feet. A 1,800-word piece on him drew 1,200 reads, and a Metz scout contacted me for the data. Sarr joined Metz in 2026 for 1.2 million euros. I kept my original assessment unchanged.

What I took from it was not that I had been right. It was that raw statistics say nothing without film to cross-check them. Sarr's numbers were excellent, but the numbers of a tempo-setter and the numbers of a safe sideways passer look identical on paper. The difference lies in the timing of the first touch, the direction of the first turn, and where a player stands when his team loses the ball.

Balmont does not manufacture stars; it only reveals who is willing to run more in order to shine.

From then on I set a fixed rule: never conclude from raw statistics alone, rewatch the footage at least three times, note the exact timestamp of every action, and cross-check at least two data sources before publishing.

Twelve criteria and a suspended season

In March 2026 European football stopped. I collected footage of 120 matches across six leagues — Ligue 1, the Premier League, La Liga, the Bundesliga, Serie A and the Eredivisie, from 2026 to 2026 — hand-coding every pressing action against twelve structural criteria: distances between lines, pressure direction, defensive angles, trigger timing and similar markers.

That dataset became my master's thesis in sports management and was later published in a French football analysis journal with 3,400 reads. It also opened the door to my job in 2026.

The greatest value of those twelve criteria is not that they produce answers. They produce a framework for stating that this data is missing and that criterion cannot be measured. After hand-coding 120 matches you know exactly where your own system is blind. A metric that cannot measure something is normal. A metric that measures it wrongly while nobody notices is dangerous.

In modern football, space does not appear on its own; it is forced open by the moving block. That holds for space on the pitch and for gaps in data.

When the midfield needs no hero

In the summer of 2026 I contributed analysis across 22 World Cup matches in Russia. France's 4-2 win over Argentina was the most discussed game, and I deliberately refused to write to the emotion of the scoreline. I logged fourteen French pressing actions in the first half, one by one, with exact minutes.

My longest essay on France's midfield trio made a point the same-day coverage missed: N'Golo Kanté was not merely cleaning up. He screened the space in front of the back four, and that screening released Paul Pogba from second-line defensive duty. Blaise Matuidi did the dirty work on the left channel, a role that barely registers in a stat sheet.

The 2026 World Cup taught me that a midfield needs no hero, only a tempo-keeper. The piece reached 5,400 reads, 4.5 times my first article. I still waited 48 hours to re-verify every figure before publishing.

The real risk is not a wrong number

There is a paradox in this trade that took me years to understand. Readers fear a wrong number. The larger risk is a report containing no wrong number at all — one that sounds reasonable from start to finish, flows well, argues coherently, and rests on no evidence whatsoever.

That kind of report is far harder to catch. A wrong number can be caught by a single cross-check. A smooth argument offers no handhold. It propagates into other people's decisions — a scout, a coach, an editor — and by the time it surfaces, it is too late.

Sitting in front of nine pages of "insufficient information", the heaviest pressure did not come from an editor. It came from me. Those three backup headlines were easy to write. I know enough about Ligue 1 to build a midfield-crisis piece without a single data point from a specific match. It would read smoothly. It would be shared. And it would be a debt.

The only way to stop myself is to question data availability before questioning tactics. Does the data actually say something different, or do I need it to say what I already believe?

A good analysis is sometimes a refusal to conclude. In an environment where everyone must have a take within thirty minutes of the final whistle, saying the sample is too small is a counter-intuitive act. It generates no argument and no shares. But it keeps the rest of the dataset trustworthy.

One detail stays with me about analytical culture in France and Vietnam. Vietnamese football is disciplined by instinct — a team runs on the conviction of its leader, and change often arrives through an emotional moment in the dressing room. French football is disciplined by structure — change arrives through a model, a spreadsheet, a technical meeting. Both have blind spots. The first lacks evidence; the second lacks the admission that evidence has limits.

What to verify next match

I did not file that piece. I sent the data desk a three-tier minimum-input checklist and a note that the nine-page report should be archived as a process incident record, not as football analysis.

For readers, the lesson applies to next weekend's fixture. When you see a stat table, look for the empty cell rather than the prettiest one. When you see a heat map, ask where that player stands when his team loses the ball. When you see a figure quoted without a source and a date, treat it as an unverified rumour.

Transfers are not a race for money; they are a race to find the right person for the right gap. That holds for players, and it holds for data.

A gap in an analytical file is not a failure. It is information. The analyst's job is to read it as it is, rather than fill it with whatever sounds most plausible.