The Empty Ledger: In Football Analysis, 'Insufficient Information' Is Also a Finding

October 28, 2026. Salt Lake Stadium, Kolkata. England beat Spain 5–2 and the U-17 World Cup ended. That evening my laptop was logging the 52nd and final match of the tournament — six venues, 52 matches, and one rule set on day one: code every goal's build-up by phase before coding it by player. Before shutting the laptop I filled the last column, which is still the first line of every report I file — "what I did not see, I do not write as seen."

Nine years later that same column returned in another form. A nine-dimension football analysis framework, handed a payload that was almost entirely empty. Article title N/A, source N/A, the list of information points blank, and the entity field instructing me to "identify from the information points above" — when no points existed above. The outcome was a single line repeated across all nine dimensions: insufficient information.

That is not the report a broadcaster asks for at nine in the evening. It is the one that still stands next year.

In football's data economy the most expensive error is not a lie. The most expensive error is a number that looks correct and carries no source. Under the Premier League's profit and sustainability rules, Everton were docked 10 points in November 2026, saw that reduced to 6 on appeal in February 2026, and lost 2 more points in April 2026 over a second breach. Nottingham Forest were docked 4 points in March 2026. Those figures travel into headlines, but they only mean something attached to a date, a rule clause and a hearing sequence. Remove the date and a number turns from news into rumour.

I learned that discipline in 2026, when all 169 goals of the Russia World Cup were logged into one spreadsheet and 43 percent of them traced back to set pieces or second balls. The whole value of that exercise sat in one qualifier — "of the 169 goals coded" — never in "modern football shows." Let me open the ledger and trace where the restart actually broke; only then does a claim get written.

Measured against that standard, the document does not hold up, and that is its most useful piece of information. Not one of the six risk families — sporting, financial, personnel, regulatory, public opinion, systemic — could be scored, because scoring risk requires at least one subject and one event. Broadcast chains, agent ecosystems, the solidarity mechanism, multi-club ownership: every pathway returned the same line, N/A.

The hardest line is not the one reading 'insufficient information.' The hardest line is 'hidden information: none derivable.' That sentence appears nine times across nine dimensions. It is not a display of humility; it is a technical decision — because on an empty input, inference stops being inference and becomes invention.

The document also recorded the conditions under which the identical framework could be re-run. Five minimum-viable inputs: article title, source and publication date; at least five information points, each with its own attribution; at least one named entity — club, player, coach or competition; an article-type classification; and any quantitative data present in the original. That checklist reads harsh, and it is exactly the line separating analysis from guesswork.

Entity-extraction failure should be treated as a hard fail, not a soft warning. Without club and player names, financial models, compliance registers and tier positioning are all useless. An 'unclassified' article type is equally serious: a match review, a transfer rumour, a financial exposé and a governance story carry materially different dimension weightings. Choosing dimensions without classification becomes arbitrary.

The glossary terms carry the same warning. Expected goals (xG) and expected goals against (xGA) are model-based estimates, not hard facts. PPDA measures pressing intensity, and a lower value means more aggressive pressing — but it only matters if you state the window and the scoreline it was measured in. Transfer amortisation, sell-on clauses, the solidarity mechanism: each asks the same question. Which fee, which contract length, which date.

The real conflict is that the industry rewards the polished report. An empty one gets read as 'the work didn't happen,' when the opposite is true — a polished report built on empty input is indistinguishable to a reader from real analysis, and that is precisely its danger. A document that looks filled across nine dimensions hands the reader confidence that a check occurred, when there was no subject to check.

The second counter-intuitive point is that this report is not a performance of humility; it is a deliverable. It carries a signal table — Stage-1 extraction health, presence of source metadata, entity coverage, classification distribution — with a trigger condition written beside each. An empty-information-point rate rising above zero is itself a signal of upstream fetch failure. That is systems analysis, and it is worth no less than match analysis.

So the gate belongs at Stage 1, not Stage 2. Source name and publication timestamp should be mandatory extraction fields, because without them the expectation-gap and rumour-credibility modules stay permanently shut. Classification should happen before analysis, not after.

I know the first question an editor asks when this lands: "So is there anything?" Yes. Our pipeline has a broken joint, and it will break again if it is not logged now. Filing late and correct has always cost me less than filing first and correcting later.

The next time a nine-dimension report arrives looking fully populated, the first question will not be how many dimensions are filled. It will be how many of them were fillable.