The Empty Report: Data Discipline in the Sports Analytics Room

At 9:12 on a Tuesday morning, the monitor in the analytics room printed nine sections. All nine returned the same line: N/A — insufficient information. No source headline. No publication. No information points. No identified entities. The person sitting in front of the screen kept both hands on the keyboard and chose not to press another key.

I read that report four times over two days. Every time, the conclusion sat at the bottom: the Stage-1 input is empty, so any conclusion built at Stage-2 would be invention. That analytics room — where I sat for years telling myself I was doing science — had just done something very few analytics rooms dare to do: it refused to answer.

There is a gap in the middle of the process, and someone left it open.

Two stages and one gap

A professional sports analytics pipeline runs through two stages. Stage-1 deconstructs the source text: title, source, article type, information points, named entities, time sensitivity, source quality. Stage-2 takes that output and expands it across nine dimensions: technical and tactical, data and form, tournament system and schedule, tour landscape and player positioning, rules and governance, team and player management, risk, media narrative and expectation, and industry transmission.

The condition for Stage-2 to work is so simple it gets ignored: Stage-1 must contain something. This time it contained nothing. No headline, no source, no type, no information points, no entities. An empty input — and under the pipeline's null-value protocol, the output must retain the nine-dimension frame with every cell marked N/A rather than fill the gap with speculation.

Put me in that seat ten years ago, at 31, and I know exactly what I would have done. I would open a new file, type “tennis,” type a few familiar names, build a table that looked plausible, and file it. The Daily Mail needed column inches and I needed column inches. The trap did not come from laziness. It came from a professional truth taught in every newsroom: a blank page is the enemy.

The summer of 2026 taught me the opposite. When COVID-19 wiped the calendar from March onward, I was temporarily out of work and sitting at home. Instead of writing about matches that never happened, I collected data from 312 Premier League, La Liga and Bundesliga matches from the 2026-20 season and split them into two groups: pre-pandemic matches with crowds, and end-of-season matches in empty stadiums. Home win rate fell from 46% to 38%. Average goals per match ticked up, from 2.67 to 2.81.

That 5,000-word analysis sat in two editors' inboxes for two weeks. Nobody replied. In week three, The Athletic wrote back one line: the most original angle of the year. They ran it as a feature. A European bookmaker then called to ask about my data source.

The lesson was not that I was right. The lesson was that I spent two weeks turning public data into something nobody else had, instead of two weeks writing something anyone could have guessed. A quiet summer turns records into orphaned numbers. People need someone to go pick them up, not another person to paint over them.

The Vietnamese market is asking the same question right now. V.League is at a stage where public data remains far thinner than in European leagues: positional data, pressing metrics and standardized expected-goals models are still rare. Platforms such as VuaBong.vn and indices such as the VangBong.vn Player Depth Index exist to fill part of that gap. But when the source data is thin, the pressure to fill the gap with guesswork grows. That is the most dangerous place in this profession.

What gets protected when a cell reads N/A

Readers routinely misread reports made entirely of N/A. They read it and think: this person produced nothing. The actual mechanics run the other way.

In 2026, at 33, I was sent to Russia as a senior analyst for the channel. For the quarter-final between Russia and Croatia, before the shootout began, I went on air with an analysis that sounded very solid: Russia had taken 45 minutes of penalty practice every day of the tournament, but Croatia had goalkeeper Danijel Subašić, who had saved three in the shootout against Denmark. I predicted Croatia would win 5-4. Croatia won 4-3.

After the match, a younger colleague texted to ask why I had not committed to a sharper number. I read that message and understood the problem: I had made a safe prediction because I was afraid of being wrong. For a month afterward, I rewatched all 64 matches of the tournament, noting every passage of play I had misjudged, then built a private spreadsheet comparing my predictions against actual results. The Russian night was scorching, and the only lesson left standing was the silence.

That silence comes in two kinds, and they are entirely different. The first is silence because you do not know — the kind I displayed on air that night, and it is cowardly. The second is silence because you know there is nothing yet to say — the kind that empty report was practicing, and it is different in kind. Silence is not the absence of an answer — it is the answer, for anyone who knows how to listen.

When a pipeline refuses to name a player who does not exist in the data, it protects three things at once. It protects readers from a fabricated entity. It protects the pipeline from destroying its own credibility on the next run. And it protects the most valuable asset in this trade — the ability to tell “not yet computable” apart from “computed.”

Nine dimensions, and the price of guessing

Imagine someone decides to fill that gap. Nine dimensions, each needing an entity. The writer picks a player. The first name that comes to mind. Now the piece has a subject, and appears to have value.

The technical dimension will assign that player a style: aggressive baseliner, counterpuncher, or serve-and-volleyer. It sounds reasonable, because every player falls into one of those boxes. The problem is that style is decided not by the writer's instinct but by point distribution across surfaces, net approach rates in key service games, and win rates in deciding games. Without that data, “style” is just a label.

The data and form dimension needs a panel: first-serve percentage, return points won, break-point conversion, winner-to-unforced-error ratio. Faking such a panel is easy, because every number looks plausible somewhere between 50% and 80%. Faking such a panel is also very hard to catch, because nobody checks each cell.

The tournament dimension needs to know whether this is a Grand Slam, a Masters 1000, an ATP 500 or an ATP 250, whether entry is mandatory, and where it sits in the calendar. Get the tier wrong and every analysis of points and pressure is skewed. The tour-landscape dimension needs the player's position among title contenders, seeds and the top-30 backbone. No name means no position.

The rules and governance dimension needs an event: a sanction, a controversy, a rule change. The team and player management dimension needs a concrete coaching relationship. The risk dimension needs an exposure subject — someone with an injury history, someone about to lose ranking points, someone in a contract dispute. The media dimension needs a labeled narrative. The industry dimension needs a triggering event.

Nine dimensions, nine different kinds of fabrication. And the frightening part is that none of them self-report. A piece that is wrong about a real player gets caught by readers within hours. A thorough piece about a player who does not exist can live online for years, because nobody has a reason to look up a name they already believe is real.

That is the real price of filling a gap with guesswork: it does not create the risk of being caught, it creates the risk of never being caught. For an analytics room, the second kind of risk is far worse.

Spreadsheets do not know what longing is

In 2026, in ESPN's analytics room, I watched footage of Josef Martínez 14 times — a 24-year-old forward who had just scored 19 goals in MLS. Instead of waiting for a superstar out of a big academy, I dug through expected-goals data and found something unusual: his no-backlift finishing style produced an abnormal conversion rate of 23.4%. I wrote a 1,200-word analysis and published it on the channel's blog. The content director called me in and said: “You have a nose for this. But stop writing like a thesis.” The following week I was given lead commentary on an Atlanta United match. Martínez scored twice; I called him “the silent predator,” and the stand laughed.

What I took from it was not that data beats the eye. What I took from it was that data only matters when it tells the story of a specific person doing a specific thing. 23.4% means nothing on its own. 23.4% belonging to a 24-year-old who had never been capped, finishing with a technique no MLS defender could read in time — that is worth writing. Numbers are seasoning. People are the meal.

In analytics rooms, people tend to have a pet: an index, a model, a player they have staked their reputation on. I have had mine. Josef Martínez was one. But the pet of the analytics room eventually has to stand on its own feet. When Atlanta changed systems, when opposing back lines learned to shut the inside channel, the 23.4% fell and the old story stopped working. The model was not wrong. The model was just old.

An empty report like the one I was reading has an advantage no model has: it can never go stale, because it never said anything.

Four tests every number must pass

Over the years I distilled four questions every number in my analysis must answer before it goes to page.

Where does it come from. A positional metric pulled from a broadcast partner's tracking cameras carries different weight from a metric pulled from a news site's aggregated stats table. If the source is unclear, the number is unusable.

How large is the sample. Josef Martínez's 23.4% conversion rate in 2026 rested on how many shots? If it is 60, that is a signal. If it is 12, that is chance packaged as a chart.

What is the comparison. A rate only means something against the league baseline. Danijel Subašić's three shootout saves sound extraordinary until you remember that a goalkeeper's career shootout sample is rarely large enough to separate from noise.

What would refute it. If I cannot imagine a scenario in which my number is wrong, I do not understand that number. This is the hardest question, and the one that has saved me most often.

These four tests are exactly why a serious pipeline needs a null-value branch. When the input fails the first test — no source — the other three have nothing to check. The only correct output is N/A.

U18 and the pressure to physicalize

In Vietnam, I follow youth age groups with a specific worry. The trend toward physicalization at U18 level is eroding the technical foundation. At 17, a player still needs thousands of hours of touches in tight spaces to control the ball in two beats. Instead, academies prioritize youth-tournament results, so they select for frame, for stamina, for whatever produces points on the weekend.

Centres such as PVF and the HAGL-JMG academy invested in measurement and sports science relatively early, and that is the right direction. But the gap between measuring and using is still wide. A young coach holding running-volume data will be tempted to use it as a selection criterion, because running volume is countable. The quality of a pass in a three-on-two counterattack is much harder to count.

This is where a decent analytics process can make a real difference. If all you have is physical data, you will always conclude in the direction of physicality. If you admit you lack technical data, you are forced to go find it — hire video coders, build your own indices, do work nobody sees. Spreadsheets do not know what longing is, and we should stop pretending otherwise.

How a fabricated number travels

There is a transmission chain I have watched many times. It starts in a piece short on data. The writer needs a number, so supplies a plausible estimate. Another site quotes it and drops the word “estimated.” A commentator reads that site on air. By the next afternoon, the number sits in three analyses, two videos and a fan-built stats table.

The cost of correction is not the number itself. The cost of correction is that nobody remembers where it came from well enough to correct it. Once a number is severed from its source, it becomes irrefutable — and in this industry, the irrefutable is the most dangerous thing there is.

Run the other way, and proper null-value handling produces a benefit few notice: it forces people to state exactly what is missing. The report I read did not say “there is nothing.” It said precisely what is missing, at which stage, and what must be supplied to re-run. That is a more useful form of feedback than a report stuffed with figures.

The contrarian angle: honesty is the floor

But I will not leave that report alone.

The way it presents itself is beautiful: protocol-compliant, no baseless speculation, structure preserved, awaiting valid data. All of it true. And all of it the floor of this profession, not the ceiling.

An analytics room that refuses to answer when it has no input has completed exactly half its duty. The other half is going to get the input. If Stage-1 is empty, the job does not end at writing N/A and closing the file. The job is to open a browser, find the original article, call the reporter, re-read the transcript, rebuild the headline, the source and the information points. A pipeline that can only say “not enough data” and cannot go find data has traded one failure for another: it has swapped the sin of fabrication for the sin of paralysis.

There is a worse version still. When “insufficient data” becomes the default answer, it becomes a shield that legitimizes laziness. People write N/A because N/A is safe and nobody can fault them for it. But readers do not need a safe analytics room. They need one willing to say: I think this is likely, and here is where I could be wrong.

In the biggest lesson I brought back from Euro 2026, when I used real-time data to say that Italy's pressing index was declining and that Federico Chiesa would be withdrawn — and Roberto Mancini pulled Chiesa at minute 65 exactly as predicted — what I remember most is not that I called it. What I remember most is my supervisor's warning afterward: do not turn yourself into a prophet, because the audience will set the bar too high. Ever since, whenever I use real-time data, I attach the limits: data cannot measure a player's psychology, and it cannot measure a sudden decision on the coaching bench.

Being honest about the limits of data and being honest about the absence of data are two different things. The first is craft. The second is the hinge.

What remains after the spreadsheet closes

That empty report will not be remembered. It has no player name, no scoreline, no controversy. It has nine analytical dimensions and one N/A repeated often enough to become a statement.

But it leaves one question behind, and I think that question will outlast any number I have ever written: if you delete this number, what is left of my analysis?

If the answer is “nothing,” then that number is carrying the whole piece — and it is not strong enough. If the answer is “a player, a decision, a moment the reader can picture,” then the number is doing its job.

Next time you read an analysis packed with numbers, try deleting them all and look at what remains. And next time you read one with no numbers at all, ask yourself whether the writer went looking — or just sat waiting for data to fall onto the desk.