An Empty Sheet Doesn't Lie: Notes from a Sports Data Writer

On Tuesday night, I opened the file and found it empty. Not empty in the sense of missing a few lines — completely empty. The title field said N/A. The source field said N/A. And the list of information points, the thing every downstream analytical step must hold on to, was an empty array with not a single element. I sat staring at the screen for about three minutes, hands still on the keyboard, waiting for a line of data to appear on its own. It never appeared.

That night I understood something that nineteen years in this trade had not fully taught me: the blank is not the scariest thing. The scariest thing is how a writer reacts to the blank.

I write about sport through data. Not the kind where you watch a match and narrate what happened, but the kind where you hunt for numbers nobody noticed, place them side by side, and try to answer a single question: what is actually happening on the pitch. That work depends entirely on one thing: the data has to arrive. When it does not, I am no longer a data journalist. I am just a man sitting in front of a white screen.

There is a line I still repeat to interns whenever they ask for the secret: Raw numbers are mud; to see the truth, you have to put your hands in it. But putting your hands in only means something when there is actually mud down there. When the bucket is empty, reaching in just gets your hands dirty without giving anyone a single extra fact.

That is why I am writing this.

How a data analysis is born

So you can picture it: every in-depth analysis my colleagues and I publish goes through a fixed chain of steps. The first step extracts information points: who, did what, when, what number, from what source. Without that step, everything after it is a house built on sand. The second step — where I stand — takes those information points and places them into a nine-dimension frame: tournament context, roster, region, finance, rules, risk, public narrative, and the transmission chain across the whole industry.

That frame does not exist to look scientific. It exists because each dimension has, at some point, saved me from a specific mistake.

The tournament-context dimension, for instance, taught me that format decides a great deal before a player even touches the ball. A double round-robin group stage is nothing like a single-elimination bracket. A league playing every three days is nothing like one with a week between games. Whether a strong team stays stable depends largely on whether they get enough recovery time, not on whether they have stars.

The roster dimension taught me that paperwork does not play football. A roster full of famous names can lose to a modest roster whose roles fit together. I once watched a team field three attacking midfielders whose natural side was the right wing, and all three got locked into the same corridor. Their mistake was not individual quality. It was that nobody agreed to play on the other flank.

The regional dimension taught me that the same team, placed in two different regions, becomes two different stories. A young talent emerging in a region with good academy infrastructure might never have emerged had he been born somewhere with no academy at all. That is not fair, but it is true, and a data writer has to say so.

The finance dimension taught me that a transfer fee is not a measure of talent. A player who has not played fifty top-flight matches yet but carries a fee of hundreds of millions of euros — that number speaks to market expectation, not to the quality of his feet. I learned I have to separate the two, or else I will keep repeating the stories of bubbles that already burst.

The rules dimension taught me that a league is shaped not only by what happens on the pitch but by what is written in the rulebook.

The risk dimension taught me that a bad outcome which has not yet occurred does not mean it does not exist. An absence of crisis signals is not the same as safety. This is the subtlest trap I have ever fallen into: seeing a clean dataset and concluding that everything is fine.

The public-narrative dimension taught me that crowd expectation can run far ahead of the data. A team that is overhyped can collapse not because it got weaker, but because of the weight of the expectation itself.

The industry-transmission dimension taught me that decisions made at the very top — by publishers, organisers, sponsors — flow all the way down to the deepest layer, which is how a supporter watches a match on a Saturday night.

All nine of those dimensions rest on the same foundation: the information points. And on Tuesday night, that foundation held nothing at all.

When the spreadsheet is empty

That empty file was the result either of a failed extraction step or of a source that was empty to begin with. From the data I had, I could not tell the two apart. And here is the crux of what I want to say to you: when you cannot tell them apart, the honest answer is “undetermined,” not a judgment that merely sounds plausible.

In every analytical frame I use, there is one non-negotiable convention. When a cell has no information, I write “insufficient information.” I do not guess. I do not fill the gap with intuition, even though my intuition about sport is not bad. This sounds simple, but it is the line between decent data journalism and a machine that produces writing which sounds intelligent while being hollow.

I have crossed that line before. In 2026, ahead of the World Cup in Russia, I was twenty-seven, working as a data journalist for The Athletic. I built a model based on xG differential and PPDA — the number of passes an opponent is allowed before your team makes a defensive action. I publicly said France would win, even though the press rated them below Germany and Spain.

My argument then was very specific. In the semi-final against Belgium, France's average PPDA was 7.8. That figure is so low it is almost counterintuitive, meaning France deliberately ceded the ball in order to counter. Belgium had a PPDA of 11.2, but their back line lacked pace. France won one-nil, and my piece was shared more than three thousand times.

I will say this plainly now, years later: Russia 2026 is where I staked my honour on the PPDA model, and I do not regret it. But I also know that success is sedative. When a frame gets one big result right, the writer starts believing the frame is right about everything. That is precisely when this trade becomes dangerous.

I remember an afternoon in 2026, Miami FC against Indy Eleven at Riccardo Silva Stadium. I was twenty-six, fresh off a master's degree, convinced that data does not lie. I carefully logged the passing of the midfielder Richie Ryan: eighty-seven touches, seventy-four passes completed, a pass accuracy of 91.9 percent. I put exactly those numbers into the piece, listed them plainly, and my editor cut it down: “Reads like toilet paper.”

I did not argue. I went back, rewatched the entire tape, and started building a frame I later called the Territorial Influence Index — combining where the ball was received, where the pass was aimed, and the space each pass opened. The second time I filed, again with Richie Ryan's numbers, but this time tied to a specific image: the turn away from pressure in midfield, then the forty-metre switch that tore the defensive line apart. My editor ran it on the front page immediately.

The lesson I keep to this day: a number only has value when it is tied to something the reader can see. The number does not replace the story. The number illuminates the story. And when there is no number at all, I have no story at all.

Silence that echoes

In 2026, when the pandemic wiped the crowds out of the stands, I was twenty-nine, working as a data editor for ESPN, covering the MLS tournament inside the Orlando quarantine zone. Empty stadiums. No fans. No home advantage. Every familiar metric suddenly warped. Possession lost its meaning. Shot counts became hard to interpret.

I collected GPS data from thirty-seven matches, measuring the running distance of every player. Players covered about nine percent less total distance than the previous season, but the number of sprints rose twelve percent. Matches were more fragmented, more explosive, and had longer dead periods. I wrote a four-thousand-two-hundred-word internal report arguing that the way we measure performance has to change when there are no fans. It was later edited into a front-page ESPN piece.

What I learned in Orlando is not in the spreadsheet. It is that I began to ask about the underlying conditions of a match before trusting any number at all. In the Orlando bubble, the data went silent, but the silence had an echo. That silence told me I was comparing two things that were no longer the same thing.

And on Tuesday night, I met another silence, one layer deeper.

The temptation of the blank page

This is the part I want to spend the most time on, because it is the most dangerous part.

When a data journalist receives an empty file but is still expected to file copy, there is a very tempting escape hatch: invent a framework. You do not need to invent actual numbers. You only need to fill the “insufficient information” cells with soft judgments that sound reasonable and are grammatically correct. Write that this tournament is in transition. That this team is showing signs of improvement. That the market is displaying cautious signals. It all sounds like analysis. Nothing in it can be caught out, because nothing in it is true.

I have seen such pieces. They are more dangerous than a piece with the wrong numbers, because a piece with the wrong numbers can be checked and corrected. An empty piece that sounds good will never be corrected, because nobody spots it. It just quietly poisons the reader's trust, one article at a time.

The rule I set for myself, after paying for it more than once: if I cannot point to a concrete image on the pitch that corresponds to a sentence in the piece, I delete the sentence. Not shorten. Delete. A four-thousand-word sports piece in which no sentence is anchored to a real passage of play, a real number, or a real moment is, no matter how smooth the prose, an empty product.

There is one specific mistake I want to tell you about, because I promised myself I would not hide it. In 2026, when the Euros were held late because of the pandemic, I was thirty, running the data desk for a European football podcast. In the semi-final between Denmark and England, I got swept up by a midfielder nobody had put on any watchlist: Mikkel Damsgaard. I calculated his pressing-recovery rate at 4.2 per match in the opposition third — the highest among under-23 players. Against England he made five tackles, all five successful, and created three chances from high pressing.

I wrote a piece along the lines of “the modern midfielder that data keeps missing.” It was shared by more than forty European outlets. Three Premier League club scouts emailed to ask for more.

But this is the part I rarely tell. I wrote that piece as if Damsgaard's numbers proved he would become a big star. The truth is those numbers only said that, at that moment, in that context, he was doing one specific thing well. I crossed the line between description and prediction without flagging what I was doing. My mistake was not in the number. My mistake was letting the number say more than it actually knew.

Since then, whenever my writing travels further than the data, I have to add a margin-of-error note. Not a reassuring phrase like “more data is needed.” But a blunt statement: if this assumption is wrong, this conclusion collapses.

A gap is a signal, not a defect

Here I want to turn in a different direction.

There is a way of looking that I learned from this trade itself: sometimes the thing that does not appear in the stats table is the most important thing. A team with sixty-five percent possession and not a single dangerous chance — that sixty-five percent says nothing about their strength. A player who runs ten kilometres less than his teammates can still be the one who decides the match, because his five runs all fell at exactly the right moments. The stats table is silent about those things. And that silence is worth reading.

Applied to Tuesday night's empty file, I can read it two ways. The first, pessimistic: this is a systemic failure, a break in the data pipeline, and it needs to be found and fixed. The second, more optimistic: this is a signal that something in the process deserves a second look before any analysis goes out.

I choose to read both, and I think that is the right way to read any dataset. An anomalous number can be an opportunity, or it can be an error. A gap is the same. A good data writer is not the one who always fills the gap. A good data writer is the one who knows when a gap is the answer and when it is the question.

In football, this shows up clearly in how you read defensive metrics. A team that lets the opponent hold the ball a lot is not necessarily defending badly. Some teams deliberately concede possession, drop deep, and wait for exactly one moment to counter. Looking at the possession table, you see them as weak. Looking at the tape, you see them playing exactly as intended. The difference between those two views is where I earn my living.

So when I say an empty file is a signal, I am not saying it to comfort myself. I am saying it because I have seen far too many overconfident analyses built on a thin data base, and I do not want to become part of that trend.

The transmission chain of a mistake

There is an aspect readers rarely think about but data writers must remember: an analysis does not end at the final full stop. It travels into news feeds, into group chats, into the small decisions of readers — into an argument online, into a moment when someone places a bet, into how a supporter sees his own club.

If the piece is right, the flaws are forgiven. If the piece is empty but believed, it plants in the reader's head a wrong model of how sport works. That is a price I do not want any colleague of mine to pay, and I do not want to pay it either.

I remember a reader once wrote to me, saying that after reading one of my analyses he decided to follow a club for an entire season. The team did not perform as predicted. He was not angry. He asked me one short question: “Did your numbers account for the club changing manager mid-season?”

I had not accounted for it. And I had to answer that I had not. He asked for nothing back. But I have never forgotten it. That is why, when my frame has an empty cell because data is missing, I write clearly that it is empty because data is missing, rather than pumping a soft assumption into it.

What the empty file taught me about the season ahead

It is the regular season now. Tables are slowly taking shape, races are still sitting quietly beneath the surface, and arguments about referees, fitness, and tactical signals have not yet become headlines. Fans are watching every match. They need to know how intense the title race is, which teams are struggling, and which signals will soon climb onto the front pages.

It is precisely in this stretch that an empty file is a valuable reminder. It reminds me that over the next four months I will encounter hundreds of spreadsheets, and most of them will have empty cells. If I fill every empty cell with intuition, I will write faster, my output will be more consistent, and readers will trust me more. But that trust would be built on sand.

I do not write to be fast. I write so that, after a season, when you scroll back through my old pieces, they still stand up. I want every prediction I make to point to a concrete chain of evidence, and every time I am wrong, I want to point to exactly which assumption collapsed — not to defend myself, but so I am wrong a little less often next time.

A forward thought

On Tuesday night, I closed the empty file without writing anything. I sent a short note to the team: source empty, need to re-run the extraction step, nothing to analyse yet. The next morning the source was restored. The spreadsheet came back. And I got to work.

But I kept the old empty file on my machine. I did not delete it. I keep it as a reminder: there are days when a writer's discipline lies not in what he manages to write, but in daring to write nothing. And in an industry where everyone is racing to publish faster, daring to stop because the data is not there may be the hardest skill of all.

If you ever read an analysis of mine and find a sentence that sounds too good, too rounded, ask me one question: which image on the pitch corresponds to that sentence. If I cannot answer immediately, that sentence deserves to be deleted. Even the prettiest one. Even one of my own.