The Empty Dossier: When Sports Data Chooses to Stay Silent

Title: The Empty Dossier: When Sports Data Chooses to Stay Silent

Hook

2:17 a.m., Liverpool. I open a scouting file a colleague sent over, expecting a few hundred rows of data on a young midfielder three Championship clubs are watching. Instead, the cursor blinks on a blank frame. No columns, no rows, no xG figure, no PPDA value, not a single line of notes. An empty dossier. Outside the window, Liverpool fog settles over the red rooftops like a shroud of steam, and I sit there asking myself: what is a data man supposed to do when the data never arrives?

There are things data never touches — like the way a stadium breathes. But an empty dossier is a different species. It is not a stadium breathing. It is the deliberate absence of breath. And in the modern sports analytics industry, that absence is more frightening than any bad number.

Context

Twenty years ago, when I was still fact-checking for a sports magazine, this trade was simpler. You had eyes, you had a notebook, you had an expert sitting beside you in the stands telling you that the number 8 runs smart but finishes poorly. Today, every club from the Premier League down to the third tier has at least one data system. They pay for event providers, for motion-tracking cameras, for transfer-valuation models. An entire industry has built up the belief that football can be measured to the last percentage point.

But that belief has a hole few people mention: the data supply chain. A metric does not generate itself. It travels through a pipeline — from camera, through recognition software, through human taggers, through computation models, and only then to my desk. A single broken link stops the entire flow. And what you receive is not a wrong number, but nothing at all.

That is what I learned in the Russian summer. In the Russian summer, silent keyboards tapped out a symphony of data. But sometimes they are simply silent, and that silence is itself a signal.

People in my trade are taught to fear two things: small sample sizes and noisy data. We are trained to chase large, clean, stable samples. But no one ever taught us what to do when that sample disappears entirely, when a whole match has not a single line of data recorded, when a club sends you a file whose name is correct and whose contents are as blank as an unwritten page.

Core

In 2026, running an xG model for a club's U23 side I was consulting for, one data row made me stop. A 17-year-old striker just back from injury had a box-touch rate more than 30% below average, yet his xG per shot reached 0.42. That figure matched none of the scouting reports. The reports said the boy was raw, inexperienced, unready. The data said otherwise.

I recommended the coaching staff bring him up to train with the first team. Many objected. They said my model was too theoretical, the sample too small, that a young player just back from injury could not be judged on a handful of shots. They were partly right. A small sample is a real problem. But what I learned afterward was not "trust the model." What I learned was: when scouting reports and data say two different things, the answer lies not with either side, but in asking why they differ.

In the friendly that followed, that player scored two goals from three shots. But I do not tell this story to boast that the data was right. I tell it because the opposite is equally true: how many times was my data right merely by luck, and I never knew?

In the 2026 season, when European football shut down for the pandemic, a Championship club contacted me to produce a report on "empty-stadium" performance. They worried that missing crowds would strangle the team's spirit. I gathered 500 matches, rebuilt the model, and found at least two things. Home teams lost only about 0.18 expected goals per match without fans — an effect far smaller than the original fear. But the surprise lay elsewhere: teams trailing behind tended to play long balls about seven minutes earlier than usual.

When the stands are empty, the numbers begin to learn how to sing. They sing of disorientation, of no longer having jeers to remind players they are losing, no longer having applause to reassure them there is still time. The coaching staff adjusted their pressing based on that data, and they took 8 of 12 points in June. I wrote the report and thought: the data saved their month. But which data? Data about what was absent — about the crowd, the noise, the presence.

Then came Qatar 2026, where I witnessed what I call the rebellion of the outsiders. Japan beat Germany and Spain by pushing their defensive line 1.2 metres higher than their opponents in the second half. I stayed up all night, reopened all my files, and asked why I had missed it. The answer was not pretty: I had focused too much on the big teams, ignoring scout data from Japan's pre-tournament friendlies. I had the data, but I did not look. Between having data and knowing how to look at it lies a gap larger than any model can fill.

And now, sitting before an empty dossier, I realise it may be the most valuable data of the day. An empty file is not the truth that the player does not exist. It is the truth that the pipeline broke, that someone failed, that a story about process is waiting to be told. An empty dataset is a reminder that every number we see could be absent, and that absence is itself data.

I am too old to believe in miracles, but young enough to know which miracles can be measured. And one of the rare miracles of this trade is the ability to tell "no problem" apart from "not yet checked." The two look identical on screen. Only discipline can tell them apart. Across 38 years of watching this industry, I have seen too many transfer decisions made on dossiers that looked complete but were really carefully painted gaps.

Contrarian

The sports analytics industry suffers from a quiet disease: it believes a complete dossier is a good dossier. In reality, a dataset so full it has no gaps left is usually a danger sign — a sign that someone stuffed, smoothed, and filled the holes with guesswork, turning unobserved dark zones into numbers that flatter the eye.

A beautiful model with every column and every row, missing not a single cell, may be lying to you discreetly. It lies in the gaps that were filled. It lies in metrics interpolated from a similar match. It lies in players the camera never captured. Correlation is not causation, and completeness is not truth — only tidiness.

I prefer a dossier with holes. It is honest. It shows me my limits. When I wrote the club report on the fanless season, I set aside a small section asking myself: what if my 500 matches represent nothing but themselves? What if that seven-minute effect is just noise? I have no answer. And it is precisely the absence of an answer that makes that report different from a bare spreadsheet.

In Qatar, I learned this through my own failure. I promised myself I would never let pre-tournament bias cloud my data eye again. But that promise guarantees nothing. Ten years from now, I may miss another team for another bias I have not yet named. An honest analyst is one who knows he will be wrong, only not yet where.

Takeaway

Every dataset is a garden — the farmer sows questions, and the harvest is contracts. But an empty garden also tells us something about next season's yield. It says the soil needs rest, the pipeline needs checking, and the grower needs reminding he is not a god.

All my life I have chased the ball, but what I am really hunting is the formula for longing. And perhaps, this season, that formula lies where the numbers refuse to appear. I will keep that empty scouting file. Not to remind myself of a failure, but to remind myself that this world still holds uncounted regions — and perhaps that is where the most interesting thing is waiting.

What I might be wrong about: I assume the absence of data is always a meaningful signal. It could simply be a dull technical glitch. But even if it is a glitch, how we respond to it is still a story worth telling.