Empty Inputs and the Cost of Unverified Sports Analysis

In July 2026, the J.League returned after nearly four months of suspension. Toyota Stadium in Nagoya opened its gates, but not a single spectator walked through them. I sat in front of three spreadsheets: fifteen years of ticketing data, final league positions and commercial revenue for Nagoya Grampus.

Each lost matchday meant an average of 14,000 absent spectators, corresponding to 1.8 million yen in evaporated ticket revenue. I built a correlation model between ticket revenue and final standing, wrote a thirty-page report and sent it to the club's communications director. No reply. Six months later, part of the proposal surfaced in the club's official campaign, uncredited.

When the stands hold not a single soul, money speaks its truest. But the lesson I kept came from the step before that one, the step almost nobody sees: I had to confirm I actually had data, rather than a spreadsheet that merely looked like data.

Sports analysis has changed enormously over the past decade. Brentford climbed from the Championship to the Premier League on a model built by owner Matthew Benham and his company Smartodds. Brighton & Hove Albion run a recruitment system tied to Tony Bloom's Starlizard. Liverpool under FSG established a dedicated research department and appointed Ian Graham as director of research. In Germany, more than half of Bundesliga clubs now employ their own analytics staff. In Japan, every J1 club hires at least one full-time analyst.

Running alongside that infrastructure is a cheaper, faster-spreading product: the report. Scouting reports, wage reports, squad-valuation reports, pre-match reports. They have structure, headings, tables, conclusions. They look like verified documents, even when there is nothing inside them.

I once received such a document on a marketing consulting project for a J2 club. Twenty-three pages. Nine sections of analysis. Every cell had text, every table had rows. Read closely, the entire conclusion section was built from three newspaper articles and a four-minute YouTube clip. No second source. No collection conditions. No signature accepting responsibility.

That was when I recognised a structural flaw in this profession. A model, a spreadsheet, an analytical framework can run to completion and return a full set of outputs while the input is empty. Nothing automatically raises a flag. Format does not lie about content, but it does not protect content either.

At the level of player data, the expected-goals model is the clearest case. A forward who scores four goals in his first eight matches will show a handsome xG figure. Second-tier English clubs, working from public data on platforms such as Understat and StatsBomb, typically hold samples of eight to twelve matches in internal scouting reports before negotiations open. Eight matches cannot separate skill from luck. The report still gets produced, still carries a "conclusion" line, and still becomes the basis for negotiation.

At the level of transfer information, the problem lies in the source chain. A typical European deal draws information from four directions: the agent, the selling club, the buying club and the press. Each has its own motive and each routinely offers a different version of the same sum. A J1 player moving to the Bundesliga might be valued at 2.5 million euros by a German outlet, 400 million yen by a Japanese one, while the agent says "no comment". Three sources, three units, three moments in time. The reader sees one headline.

At the level of valuation, a midfielder listed at 25 million euros on Transfermarkt does not mean a club will pay 25 million euros. That is a community-maintained reference index, not a transaction. When such an index slips into internal financial reporting without a methodological warning, it becomes a false fact.

My dual German-Japanese market position allows a concrete comparison. The Bundesliga operates the 50+1 rule, meaning supporters retain voting control at most clubs, and broadcast money is distributed through a central model. The J.League sells rights as a single package and allocates revenue by final ranking plus contribution metrics. A player-valuation model built on the assumption that "clubs sell because they need cash" works in England but means nothing in Japan, where owners are typically large corporations — Toyota at Nagoya Grampus — and the objective is not transfer profit.

A data table does not know how to lie, but whoever reads it must know how to listen.

What these cases share is a procedural error. The analyst starts from the framework, not from the source. The framework is always ready. The source is not. When the framework runs first, every empty cell tends to be filled with the most plausible thing available, and the most plausible thing in sports analysis is usually whatever is already in print.

I set one rule for myself, and it costs more time than any other: three independent sources for every numerical claim. Independent means not the same agent, not the same newsroom, not the same original article. If there are not three, the cell stays empty with the reason stated. The report looks worse. It is also more correct.

I started with a blog in the Tokai region and learned that truth needs an address, not a reputation. In 2026 I analysed twelve matches of the young forward Riki Matsuda and predicted Nagoya Grampus would be relegated unless they switched from a 4-4-2 to a 3-5-2. The piece drew 140 reads. Every data point in it had a source, a date and a match attached. Those first fourteen readers could verify every line.

The paradox sits here: the analyst who refuses to speculate is usually punished by the market.

Newsrooms need copy on deadline. Recruitment departments need a report before the transfer window shuts. Investors need an index to act on. In all three situations, a report stating "insufficient data" is commercially useless, however technically correct. Whoever writes it will not get the next assignment.

The market pays for confidence, not for accuracy. A forecasting model with twelve scenarios and explicit probabilities sells. A single page reading "no conclusion possible" does not.

I do not think the answer is rigid refusal of everything. Sports analysis still has to appear before the match is played, and before the match is played there is always insufficient data. The difference between a disciplined forecast and an empty statement is that a disciplined forecast states what it assumes, how large its sample is, and how it will fail if that assumption breaks. An empty statement only delivers a result.

Every market shock has had its shadow drawn three years in advance — if you are willing to look into the gap.

The gap is rarely in the data. It sits where people forget to ask where the data came from.

What I carry from the summer of 2026 in Nagoya is a small habit: whenever an analytical table looks too complete, I go looking for its empty cell first. The empty cell says more than the filled one.

The sports industry is producing more analysis than at any point in its history, and most of it has never been traced back to a source. Fans now read data tables as closely as they read commentary. Reading data, therefore, is becoming part of watching football — not in order to believe, but in order to know which parts deserve belief.