The Night the Spreadsheet Came Back Empty: When Football Loses Its Own Evidence

Two minutes past two in the morning in Manchester, rain on the window, nothing in the room but laptop light. I opened the file. The columns sat exactly where they should — minutes, sprint distance, passes, pressing triggers, home or away. The header was immaculate. Underneath it there was not a single row.

No error message. No warning flag. The system reported the job as complete.

I pushed the chair back. I know this feeling. In May 2026, furloughed in Farlow, I sat through four hundred archived matches behind closed doors and kept arriving at the same thought: the thing that is missing is the thing that speaks loudest. That night a report landed on my desk with nine sections and nine tables, and one sentence repeated in every cell — insufficient information. Not one number had been invented. That was the most honest part of the document.

Football journalism is a pipeline now, not a column. A club's analysis department, a broadcaster's graphics desk, a newsletter writer with a spreadsheet — all of them work in two stages. The first stage pulls information points out of a match, a report, a feed: who played how many minutes, what changed at which minute, who drifted into which half-space. The second stage runs the analytical frame over those points — tactics, money, rules, risk, the credibility of a rumour. If the first stage returns nothing, the second is not weak. It is impossible. That is not a limit of skill. It is a shortage of material.

In 2026 Nathan Croft lost his Saturday radio slot in a new-media reshuffle and stopped writing match reports in favour of system audits. Manchester City finished that season on 100 points and 106 goals, and he logged every Kyle Walker and Fabian Delph inversion across all 38 league matches, alongside 412 third-man runs. I learned to work that way. I counted every inversion myself with the television paused, not in a notebook but on the screen. I first saw the inverted full-back not as a tactic, but as a confession — it only holds if both centre-backs can play.

My error was one of timing. I argued in November that the system was unrepeatable without two ball-playing centre-backs. Right about the mechanism, wrong about the timeline. What I had misread was patience: the structure does not change, the people inside it do.

At Russia 2026 I ignored the favourites and followed Croatia through three consecutive extra-time knockouts. I logged Luka Modric's 694 tournament minutes separately, in a different tab. England scored nine of their twelve goals from set pieces and still went out in the semi-final to Mandzukic in the 109th minute. I missed my return flight re-watching the Japan-Belgium tape and bought a new ticket myself. That was the first time my data, not my voice, carried the story.

When the pandemic shut the sport down in March 2026 my commentary contracts were suspended and I was furloughed for eleven weeks. I watched four hundred archived matches, learned Python, and when the Bundesliga restarted on 16 May I pulled every ghost game of the season. The home win rate had fallen from 43 per cent to 33, with away sides outrunning hosts by roughly 1.4 kilometres per match. I published it as 'The Crowd Was the Trigger.' In empty stadiums, I heard the game — and what I heard was that home advantage is manufactured by throats, not by turf.

Back to the file. Nine sections, nine tables, the same answer in every cell. No club, no player, no competition, no timestamp, no source. One thing survived: the domain label, football. It is the equivalent of a team sheet with eleven names and no positions. The sheet is legal. The match is impossible.

A bad article and an absent article are not the same object. A thin piece still yields two to five information points — who played, how long, in what competition, according to whom. Zero points means the text never arrived. The scrape stalled behind a paywall, a JavaScript wall, a geo-block. Or the wrong payload was routed: a URL, a video embed, a PDF instead of prose. Or the encoding broke. Or the CMS was still holding a draft. Or the domain tag was applied by default to something that was never football at all.

The first explanation is the likeliest, and the reason is habit rather than statistics. Major publishers now hide their body copy behind at least one layer of JavaScript, so the scraper pulls and finds white space. Returning white space is not a scraper failing. It is a scraper refusing to invent.

The real danger is not the empty return. It is the quietly wrong one. Suppose a club's data department loses one match out of a 38-game feed. The model now counts 411 third-man runs instead of 412. No alarm fires, no error band widens, one digit drops by one. The journalist writes it, the broadcaster puts it on a graphic, and nobody knows a match never arrived. The most dangerous analytical failure is not the one that shouts. It is the one with a valid schema.

I have made that mistake myself. Through 2026-18 I published my raw spreadsheets, which made the work hard to copy and my deadlines impossible — I missed four that season. Every number cost me minutes counted by hand, and counting minutes never ends. But the habit taught me that an empty cell is a question. It asks: do you not know, or do you not want to know?

Verification needs a threshold, not a treadmill. Right now I have an empty file, a valid schema and one surviving label. I could ask for more clips, more data, more sources. What is absent will not appear because I ask again. So the threshold is: enough to support the thesis, and no more. The thesis here is simple. The information did not arrive, and we need to know why.

The pipeline needs a hard gate. A payload with fewer than three information points, an empty entity list, or no named source should never reach the second stage. Send it back to ingestion with an error code. And if a report does slip out in that state, stamp it plainly: blocked, not analysed. An empty framework that reaches a decision layer is not a data gap. It is an editorial breach of trust.

The damage is sharpest in a transfer window. Source tier is the only instrument for grading a rumour — who wrote it, a reporter, a tabloid, or the club's own channel. With that cell blank, a fee figure tells you nothing about whether it is a negotiation, an estimate, or an agent's balloon. Nine immaculate sections cannot rescue one empty source field. Everything ends up weighted at zero.

There is an asymmetry nobody in football discusses. Clubs with a data department see an empty feed, trigger an alarm, open an investigation, replace the scraper, admit the fault. Clubs without one receive nothing at all — and name that nothing instinct, mentality, a winning habit. Same event, two explanations. For the rich club it is a bug; for the poor club it is character. Constraint economics does not always count the information gap. Sometimes it converts the gap into a virtue.

Now the reverse. Everyone assumes more data means better analysis. In my experience the real failure of modern analysis is not a shortage of data but an excess of confidence in it. A feed that never returns empty cannot be caught returning wrong. A null payload is a gift in that sense. It is visible. It admits the gap. A report that records its own incompleteness cannot deceive its reader, and that is the opposite of corruption.

The strongest pressure in this trade is the pressure to fill the template. Nine tables in front of you and the hands itch: put a name in, write a fee, guess a probability. Television punditry does it nightly, installing desire and hunger wherever evidence is missing. I would rather publish a visibly incomplete report, because a reader can see incompleteness. A reader cannot see manufactured completeness.

One space must stay empty — the space outside the model. Deflections, panic, a moment of genius. Nobody's expected-goals model designed Mandzukic's 109th-minute goal. The replay never explains; it only interrupts the argument we were enjoying. An analyst who plugs every gap with a model ends up describing a game nobody plays.

And one more thing. Esports did not borrow football. Football now describes itself in the language of esports — tracking data on every pass, load models on every sprint, probability on every decision. That is progress, on one condition. The game is writing its public diary in a language it hopes nobody reads. The empty file is a page from that diary, with a date and a headline and nothing underneath.

The variable for next season is not a formation or a star's fitness. It is who audits the pipe. Who notices that a feed silently lost a match, and which report absorbed the loss? The club that can answer that will hold an edge in the transfer market, because its numbers will be worth believing.

The 120th minute does not ask who is fit; it asks who is still honest. The empty spreadsheet asks the same question before kickoff. When the schema is immaculate and the cells are bare, you have to choose — the structure, or your own eyes.