Empty Data: When Sports Analysis Admits It Knows Nothing

A February morning in Turin. I opened the newsroom inbox and found a structured report. Seventeen fields. Every one returned the same value: N/A. Information points: zero. Core viewpoints: zero. Identified entities: zero. The risk section was full, but every entry described the pipeline itself rather than any team, driver or championship.

The editor asked me one question: "So what do we write?"

The honest answer was: nothing. At nine in the morning in a sports newsroom, that is not an answer anyone wants to hear. The problem of what to fill a gap with when the system declares it knows nothing has become the defining problem of sports analysis over the past decade.

On 13 November 2026, also in Turin, I replayed the Italy versus Sweden World Cup play-off second leg for the eleventh time. I drew fourteen pressing diagrams, annotated every minute, and marked the dead zones between midfield and defence in the 4-2-4 that head coach Gian Piero Ventura had built. I filed the piece with data attached. An editor twelve years my senior dismissed it in half a sentence: girls write about tactics for decoration. The article ran only after I gave him enough evidence that he had no reason left to refuse.

I bring that up because it is a miniature of what is happening now. In 2026, people distrusted data out of prejudice. In 2026, people trust data out of habit. Both extremes are equally dangerous, and both begin in the same place: nobody checks whether the data actually exists.

The infrastructure changed. The discipline did not.

Fifteen years ago, a sports desk ran on notebooks and memory. An F1 reporter finished a race at Monza with pit times written in longhand, and he was the only data source the newsroom had. Today, every race weekend emits millions of telemetry points: brake temperatures, engine maps, per-lap fuel consumption, steering angle, suspension load. Every Serie A matchday emits thousands of spatial and temporal events.

Between raw telemetry and a readable analysis, however, sits an intermediate layer few outsiders ever see: the extraction layer. This is where text, images, audio and tables are broken down into structured information points, tagged with entities and timestamps, before they move up to the interpretation layer. When extraction returns empty, interpretation has two options: stop, or invent.

The sports industry chooses the second far more often than the first. Not because anyone wants to deceive readers, but because publishing schedules do not permit a blank space to survive half a day.

I once sat beside the analysis unit of a Serie A club on a rainy afternoon. Three screens: match replay, event data table, handwritten notes. When the event table drifted forty seconds behind real time in the second half, nobody in the room noticed. They kept reading, kept commenting, kept drawing arrows. Forty seconds equals roughly three transition situations. All three vanished from the analysis, and no one could trace them afterwards.

Three ways a data pipeline breaks

When a report returns nothing but empty values, the cause is almost always one of three types.

The most visible is a source document that is empty or malformed: a file that will not open, broken character encoding, a headline outside the read zone, a table pushed into an image with no text layer.

The harder one to spot is a misconfigured extraction logic. The document contains plenty of facts, but the rule set searches the wrong field, the wrong number format, the wrong source language. An Italian article read by English rules returns almost nothing. A timing table using decimal commas read by decimal points returns meaningless values, or nothing at all.

The third is the type analysts rarely admit to: the source genuinely contains none of the information being sought. The item may be an administrative notice, a fixture list, a short social post, or an emotional column carrying no facts whatsoever.

All three produce the same blank screen but demand three different responses. For the first, fix the source. For the second, fix the rule set. For the third, fix the expectation. Confusing these three has produced most of the bad sports analysis of the past decade.

I have seen both errors. The first: a newsroom treats an empty source as a broken pipeline, reruns it until something appears, then settles for a lower-quality source to fill the hole. The second: a newsroom treats a broken pipeline as an empty source, concludes nothing happened, and publishes the silence as a finding.

What the racetrack taught me about blank space

In Formula 1, I learned a rule football analytics often forgets. When a telemetry channel goes dark, engineers do not conclude the car stopped running. They conclude the sensor failed. The distance between those two conclusions is the entire foundation of measurement discipline, and it applies unchanged to football.

A match with few blocked shots in the event table is routinely described as a controlled game. The real cause may be that the tracking system lost three players after half-time as the light changed. A match with an unusually high pass count is routinely described as excellent circulation. The real cause may be three long stoppages and an automated system filing every safe sideways pass into one bucket.

Based on my own experience watching Serie A matches, I would argue that most of the tactical labels readers consume weekly are written from data that has never been cross-checked. Nobody intends to be wrong. Nobody simply asks about the extraction layer, because that layer is invisible.

There are 22 players on the pitch, but the match is really played between two brains. The first brain reads the game. The second reads the machine that is reading the game. In the current era, the second is where matches are lost.

Atalanta, 98 goals, and the value of a self-built dataset

In 2026, when the pandemic stopped football, I built a pressing dataset for Atalanta under Gian Piero Gasperini from the 2026-19 and 2026-20 seasons. I logged 98 of their Serie A goals to find transition patterns. I used no aggregated feed. I broke down each goal myself: minute, ball recovery zone, number of passes before the shot, position of the last player to touch the ball before the assist. Four months, mostly at night, when the house was quiet and the Turin connection was not congested.

When football returned to empty stadiums, I wrote a piece based on 120 matches showing that home sides lost roughly 15 percent of their pressing pressure without a crowd. A well-known analyst shared it, and it drew 50,000 reads.

What I learned from that process was not in the conclusion. It was this: had my extraction layer failed and I never noticed, I would have had no 98 goals to talk about. I would have had an empty spreadsheet, and I would have had to write about that emptiness — or worse, write about an imagined Atalanta, pressing convincingly inside a faulty table.

The grey zone is not where the light is missing. It is where football is most real.

In 2026, at the World Cup in Russia, I wrote a long analysis of how Isco moved into the gaps between lines during Spain's 3-3 draw with Portugal. The piece was cut in half for the usual reason: readers do not go that deep. I learned to put the main argument in the opening paragraph and attach hand-drawn diagrams. After three revisions, the core section on Spain's central rotation survived intact. The real lesson sat elsewhere: if a diagram places one movement a beat too early, the entire argument collapses. The error is not in the prose. It is in the note-taking layer underneath.

The temptation to fill the gap

The strongest objection to my argument does not come from engineers. It comes from editors, and it is entirely reasonable.

An empty space on a page is not honesty, the argument runs, it is a missed opportunity. If the transfer window is hot, readers need the mechanics of negotiation, contract structures, academy systems, financial fairness rules. With no new facts, a journalist still has a duty to write about context, history, the cost of waiting.

I agree with most of that. Writing about absence is a legitimate genre, and in many weeks it is the only honest one. What I object to is a specific move: writing about a blank space while presenting it as though it contained content. That is the difference between a piece that states clearly we have no facts yet, and a piece that disguises the lack of facts as analysis.

The transfer market is where this disguise thrives. Every new contract is a hypothesis. The match is the experiment. But a rumour system never returns empty, because its input is not data but desire. Its extraction layer always works; it simply extracts from people.

My World Cup theorem does not predict the champion. It predicts who collapses first. Applied to sports media itself, the pattern is clear: an organisation without a verification gate collapses at the extraction layer first, and it collapses silently, because nobody audits a layer they cannot see.

There is a subtler temptation reserved for analysts: over-modelling. When data is thin, the instinct of anyone with an engineering background is to build a model to fill the space, turning a handful of points into a law. I have done this, and I have been wrong. After every model, I force myself to write a counterexample. If I cannot think of one, the model is not ready to publish.

The verification gate

Back to the report with seventeen empty fields on my desk.

The task is not to write an analysis of what is absent. The task is to place a verification gate at the extraction layer: a mandatory checkpoint where the system answers three questions before any content moves forward. How many information points were extracted? What is the timestamp of the source? If the count is zero, which type of breakage is it?

That gate is not an engineering obstacle. It is an editorial decision written as code.

In the coming season there will be thousands of such blanks: a race postponed by rain, a grand prix red-flagged mid-distance, a fixture rescheduled, a transfer window closing without a single signature. The organisations that handle them well will not be the ones with the most data. They will be the ones with the discipline to say, on the record: we do not know, and here is the layer that failed.

I do not believe in trophies. I believe in the system that operates to produce trophies. Inside a well-run system, a blank space is not a failure. It is a decision waiting to be made.

An empty stadium is not an anomaly. An empty stadium is an operating theatre.

What I want to know this season is not who wins. What I want to know is which newsroom, which analysis unit, will be the first to state publicly that its data pipeline broke — and where. Whoever does that first will win the only thing worth winning over the next ten years: the right to be believed.

Esports taught me that the meta always shifts. Football does too, just one beat slower. That beat currently belongs to the people who check the data before trusting it.