The Empty Report: When the Sports Data Industry Sells You False Precision

09:12, Tuesday. A fourteen-page PDF sat in my work inbox, sent by a sports data provider my newsroom pays on an annual contract. The cover had a logo, a report code, and a Methodology section in proper italics. Inside were nine analytical dimensions, each with its own table. In all nine tables, every single cell carried the same line: Insufficient information to assess.

I read all fourteen pages. No league, no team, no player, no match date, no metric. Nine analytical dimensions about something that does not exist.

Then I turned to the last page, the Risk Warning section. One cell there had been filled in: medium risk level. Probability: medium. Impact: medium. Recommendation: re-run the data ingest.

That picture says almost everything about the sports data industry I have worked alongside: a machine that could not find a single basketball event, yet still stamped that emptiness as medium risk.

I kept the PDF. It sits in a folder called Process Evidence. A report with no data is still data — if you know how to read the blank cell.

People look at the score. I look at who gets paid after that score. And here, the only thing delivered was a cover page.

For seventeen years I have sat on both sides of this table. I have been an analyst's assistant, built my own tracking sheets game by game, kept a data journal for every round because editors would not approve notes that looked too small. I have also been the one ordering these report packages, approving the invoice, then explaining to the managing editor why a fourteen-page document contained no names.

The sports analytics market runs on a simple principle: the billable unit is the page, not the finding. A nine-dimension template can be replicated across hundreds of matches, thousands of players, dozens of leagues. A human analyst cannot. So the product shipped to market is the frame, and the content is whatever has to be poured in to fill it.

Vietnamese newsrooms typically subscribe to three to five data feeds like this. During the transfer window volume triples or quadruples, because reader demand for squad and contract news spikes. That is precisely when the share of empty reports rises fastest — and precisely when almost nobody checks.

To see why this matters, remember that professional football is an industry that leaves dense paper trails. UEFA's financial rules cap squad cost at 70 per cent of revenue from the 2026-26 season. The English Premier League limits losses to 105 million pounds across three rolling years. In Spain, release clauses are legally mandatory and publicly registered. Every large number has a document behind it. Every contract has two pages: one public, one real.

Which means when a report talks about money without citing a single document, that is a signal. When a report cannot cite even a player's name, it is a much stronger signal.

A blank cell in an analytics report is not always the same kind of blank. There are three kinds, and they behave very differently.

The first is the honest blank. The ingest worked, the facts were extracted, but they do not support a conclusion. A pure rules explainer, for example, will have no player efficiency metrics, and saying insufficient information on the player dimension is the correct answer. This kind has a clear fingerprint: the other dimensions still carry content. The article has a title, a source, a date.

The second is the defensive blank. The provider fills it in to limit liability. They know the information, but it is not verified well enough to carry their name. This kind usually comes wrapped in phrases like a source close to the situation, is believed to be, could well. That is evasive writing, not false writing.

The third is the cover blank. The pipeline broke at ingest. No source article, no title, no source, no timestamp. But the template still has to ship on time, so all nine dimensions are flooded with the same sentence. This kind has a tell: it covers every dimension at once, including dimensions that cannot naturally all be empty together. A rules article can lack player data. But no basketball article lacks a league, a team, a player, a date, and a source all at once.

I found it in a data table nobody looks at — the Risk Warning table. It was the only place in that fourteen-page file where text had been typed in. And it was filled in with the word medium, with no explanation attached.

This is the methodological crux. When a system lacks sufficient information across all nine dimensions, it cannot simultaneously have enough information to score risk. Those two conclusions exclude each other. Their coexistence on one sheet proves the risk section was not calculated. It was typed in to fill space.

False precision is a product, not a bug. It is designed to look like real precision: same font, same table layout, same notation system. A reader skimming it sees a professional document. Only a reader who reaches the last page sees the contradiction.

During the transfer window, the flow of this material settles into four fairly stable steps.

Step one, an unidentified source emits a fragment. It could be an agent, a club staffer, or just a social media account. Nobody records the source.

Step two, a foreign aggregator reposts it with the words reportedly or believed to be. That site has traffic, formal credibility, and no responsibility for the original content.

Step three, a domestic newsroom translates it and attributes it with the formula English media report. At this point the fragment has been dressed in legal clothing.

Step four, the fan community turns it into fact. A club name gets attached to a number. That number starts being cited back as verified reality.

Four steps, and not one of them requires an original document. I do not believe testimony. I believe the fingerprint on the contract and the shoe print in the corridor. In this case, neither exists.

There is a tool that anti-doping uses against exactly this kind of drift, and the sports data industry should study it. An athlete's biological passport does not look for a banned substance in a urine sample. It tracks one person's biological markers across years and raises the alarm only when the curve deviates from itself. A single sample proves nothing. A whole sequence of deviation is the signal.

Applied here: one report with one blank cell is not alarming. But nine blank cells at once, repeated across many reports, in the same time window, from the same provider — that is a curve deviating from itself. It is a sign the pipeline is broken, not a sign the market is quiet.

Based on my experience following games across many seasons, I have settled on a simple reading rule. I do not read the conclusion first. I read the methodology section first, because it is the only place in a report where the author is forced to be honest about what they actually did. If the methodology section has no field for ingest status — no place to say whether a source article existed — then everything after it is decoration.

A decent report should answer three questions before answering any tactical question at all. First, what event is being discussed, on what date, published by whom. Second, which tier each claim belongs to: explicitly documented, reasonable inference, or highly speculative. Third, if data is missing, where it is missing and what would be needed to fill it. Those three questions cost far less than a nine-dimension table.

Scandals do not fall from the sky. They are initialled, scheduled, and staged step by step. An empty report is the same. Nobody types the word medium into a risk cell at three in the morning out of mood. It happens because the process allows it, because the deadline demands it, and because almost nobody reads that far.

I do not want this piece read as an indictment of every number. Doing that would make me catch the exact disease I am pointing at.

An honest blank is a professional quality, not a defect. In many cases the most accurate answer an analyst can give is: not enough data to conclude. Saying that requires more nerve than inventing a figure. Many analysts I respect say it every week, and lose work because of it.

The problem is that an honest blank is trivially easy to counterfeit. Add a cover, add a logo, add a methodology section, add a risk table filled in with the word medium. From the outside, an honest document and a broken one look identical. Only the last page distinguishes them. And in today's publishing industry, almost nobody reaches the last page.

It is also fair to name the demand side. Fans want definitive answers. Editors want copy on air on time. Distribution platforms reward frequency, not certainty. A piece saying there is no evidence yet will never draw the readership of a piece saying the deal is progressing well. That pressure does not come from data vendors. It comes from us.

I once had three months of a suspended football summer to read the sponsorship filings of a major club. I traced the money through six intermediary entities using public corporate registries. The piece drew three legal threat letters and no lawsuit. What I learned was not a tracing technique. What I learned was that the things most worth saying usually sit deep in an appendix, on a line nobody is paid to read.

A risk table filled in with the word medium while all nine dimensions above it are blank is that kind of appendix. It is small. It is boring. And it says far more than the cover page.

If you are a reader of transfer news this window, there are three checks that take under two minutes. Look for whether the piece names an original document, whether it names a publication date, and whether it distinguishes documentation from inference. A piece that fails all three right now deserves to be read one beat slower.

If you work in the trade, the task is shorter. Do not fill the blank. Publish it, with the pipeline status and the reason the data is missing. A half-page document stating that the source article failed to ingest is more honest than a fourteen-page document that says insufficient information nine times and then scores risk on the last page.

I still keep that PDF. Not as evidence against anyone. I keep it because it is the most accurate description of something this industry has not yet named properly: a machine that has learned to look busy without needing to know anything.

And the question I have not answered in seventeen years is not how to fix that machine. It is this: if tomorrow every empty report were returned with a one-line request to ingest, how many newsrooms would still have enough copy to fill the page?