The Data Consultant and the Empty Report: The Limits of Numbers in Professional Sport

On a Tuesday morning in Surabaya, rain hammered the tin roof of a small office on Darmo street. I reopened a report I had just sent to the coaching staff of an Indonesian first-division club. Fourteen pages. All nine analytical cells returned the same sentence: insufficient information to assess.

No tournament name. No ranking points. No timestamp. No player names. Not a single line of results. The analyst who uploaded the source file was twenty-seven. He messaged me that the retrieval system had failed and nobody managed to correct it before the deadline.

A small incident in an ordinary week of Indonesian football. But sitting with those fourteen blank pages, I realised I was looking at my own profession from a position it rarely gets to occupy: the position of having nothing to say.

Across more than two decades of watching and analysing sport, I had grown used to the numbers always having something to tell me. Sometimes I read results, sometimes advanced metrics, sometimes the origin point of a shot, sometimes simply the distance between the lines when my team lost the ball. That empty report taught me something different: the hardest part of analysis is not finding the right number, but recognising when no number is right at all.

Nine windows and the ritual of the trade

The broken report was not the product of carelessness. It came out of a nine-dimension analytical template that two colleagues and I built in 2026, originally for scouting at a youth academy in East Java, later adopted by clubs and a couple of federations on trial.

The nine dimensions: technical and tactical analysis; player form and data; tournament system analysis; global landscape and team positioning; rules and institutions; coaching staff and support systems; risk surfaces; public narrative and expectations; and finally industry-level transmission, from equipment brands and tournament commerce to regional markets, talent pipelines, and capital.

The idea behind it is simple. One player performing well in one match says nothing about next season. One team winning one title says nothing about that country's development system. To read correctly, an analyst must open several windows at once and check whether they look in the same direction.

But the nine-dimension template has a fatal weakness I only recognised when it returned all zeros: the more detailed an analytical framework becomes, the more it creates the illusion that it always has an answer. A beautiful structure makes you forget a rough truth. If the input is empty, the output is empty, and every blank cell is the most honest confession the system can make.

Based on my experience watching matches, younger colleagues usually make the same mistake: when they hit a blank cell, they fill it with speculation, with instinct, with a story that sounds plausible. I understand why. Nobody wants to submit a fourteen-page report consisting of the words "insufficient information". The coaching staff will ask where the money went. But that is precisely why this profession contains a large share of analysis that sounds highly professional, is beautifully presented, and has no basis beyond the author's fear of leaving a cell empty.

I belonged to that group once. And I paid for it.

The 2026 play-off: the model returned 1.8, reality returned zero

In 2026, at thirty-six, I was a data consultant for Persebaya Surabaya in Indonesia's Liga 2. The team entered a promotion play-off against PSIS Semarang. I prepared a dossier of nearly forty pages, and the section I trusted most was the expected-goals model.

My model predicted Persebaya would generate 1.8 xG if they executed the plan. On that basis I recommended pushing the defensive line high, committing more players into the opposition half, and trusting that pressure would produce an early goal. The result: a 0-2 defeat.

Watching the footage back, I saw clearly what my model had obscured. PSIS deliberately conceded territory, dropped their back line deep, closed the central lane, and invited Persebaya to shoot from distance. Persebaya took nineteen shots; fifteen originated from outside the box, most from narrow angles with nobody arriving for second balls. The total xG was not wrong arithmetically. It was wrong narratively.

I had ignored two things. First, PPDA — passes allowed per defensive action — which showed PSIS had not lost control at all but was actively choosing where to defend. Second, shot origin, which every model calculates but very few people actually read before making a decision.

That error gave me the first line of a professional code I still reread every Monday morning: "The model was not wrong. I was wrong to make it speak in place of my own eyes." An aggregate metric succeeds by compressing information. It fails the moment its user forgets that compression means loss.

From the following season I changed the process. Every number sent to the coaching staff had to carry three things: match context, the zone in which it originated, and a warning line about the conditions that could strip it of value. If there was not enough data to describe the context, I wrote plainly that the section had no basis for a conclusion. At first people complained. Later, those warning lines became the most carefully read part of the report before kick-off.

Croatia 2026 and the number nobody picks up

In 2026, at thirty-seven, I followed the World Cup in Russia and wrote a series of analytical pieces for a local sports outlet serving readers in Indonesia and Vietnam.

While studying Croatia, I noticed a detail the mainstream media barely mentioned. Croatia's PPDA in the group stage was only 9.2 — not the highest pressing figure in the tournament, and read on its own it suggests a team sitting back and waiting. Splitting the data by pitch zone reversed the picture: Croatia recorded the tournament's highest rate of ball recoveries in the opposition half, 12.4 per match.

Croatia did not press continuously. They pressed at precisely the right moments. Luka Modric and Ivan Rakitic did not run more than everyone else. They chose when to run better than everyone else. A player's true value lies in where he runs and in the moment he stops.

I wrote that piece with a headline built around the finding rather than a generic description, and it was shared more than two thousand times across regional tactical communities. That was the second important lesson: a number everyone ignores is often worth more than a number everyone quotes, because the common number has already been fully priced into the shared story.

But I also remind myself how easily that success becomes a trap. "Croatia did not win the trophy, but they showed me a truth hidden inside a number." They lost the final to France. The number I found did not make them champions, and I am not permitted to write as though it did.

2026: when data itself became afraid

In 2026, the pandemic stopped almost every football competition. At thirty-nine, I was a data consultant for a first-division club, retained through the lockdown to prepare for football's return.

The board asked me to forecast form once the league restarted. I used data from the first fifteen rounds, built a fairly careful model, and recommended the team continue its possession-based approach, since that foundation had produced results before the stoppage. The team lost three straight matches when play resumed.

The causes lay in variables my model did not contain. Empty stadiums removed the crowd pressure that normally slows an opponent's pressing tempo. Counter-attacking sides pushed higher and pressed harder, and my team repeatedly lost the ball in their own half within the opening twenty minutes of each half. Our lines stretched apart precisely because we trusted our ball control too much.

"The pandemic taught me that data becomes afraid too — when the world stops, numbers mean nothing." That reads like a slogan, but it is an accurate technical description. My model was built for a world with crowds, with normal routines, with a stable calendar. When that world vanished, the model did not degrade gradually. It lost validity instantly, like a map of a city whose streets have been redrawn.

I wrote the piece "Data Speaks, But It Must Also Listen" and changed my working method from then on. Every analysis had to present at least two scenarios, never a single confident claim. I began tracking non-traditional metrics such as the average distance between lines in crowdless conditions, and annotating clearly that such metrics hold value only in the exact context in which they were measured.

The 2026 failure also taught me something about writing. Readers do not need an expert who is always right. They need an expert willing to state the limits of his own work. The piece admitting my model's collapse was read more widely than every confident analysis I had published before it.

Italy 2026: reading empty space instead of the ball

In 2026, at forty, I approached the European Championship differently. Instead of starting from PPDA, I started from a question about space: where does this team create gaps, and how?

Roberto Mancini's Italy was my deepest case study. My conclusion was that Italy did not control matches by pressing continuously. They controlled them by compressing space horizontally, keeping the lines close so opponents had no cushion, then suddenly switching play to the opposite flank. Positional data showed Italy had the tournament's highest rate of switches to the mirrored wing, 18.3 per match.

That figure does not say Italy passed a lot. It says Italy knew how to pass long in order to open space, and knew to pass at the exact moment an opponent's block had just shifted. It is a form of compression and expansion quite different from pressing.

I predicted Italy would reach the final from the group stage, and the analysis was later published in an international data journal. But what I kept for myself was not the correct prediction. It was the change in how I watched: I began tracking where the ball was not, rather than only where it was. "Numbers are the prayer, but instinct is the candle — I light both whenever I read a match."

Since then, every report I send to a club includes a heat map I draw by hand, because the software is not flexible enough to show what I want to show. The coaching staff occasionally complain it is hard to read. But those hand-drawn maps are what let coaches see the gaps the tables cannot transmit.

VAR: the noise moves rather than shrinks

Working with competitions across the region gave me a chance to observe how VAR operates at different levels, from leagues with fully equipped video rooms to leagues with only a few camera angles.

What I have observed across several seasons is this: controversy does not disappear when VAR arrives. It relocates. Previously it happened on the pitch, between referee and players, and ended with the final whistle. Now it happens in the review room, in post-match technical meetings, and on social media for days afterwards.

The reason is concrete. VAR does not remove the grey zone of the laws. It moves that grey zone from a referee's instantaneous decision to a multi-step process: choosing the frame, choosing the moment, choosing the camera angle, interpreting the intervention threshold. Every step is a human decision. If a frame is selected a twentieth of a second early or late, a verdict on a goal can flip — and nobody calls that a technical error.

In data work we have a principle: when a statistical system grows more complex, it is not necessarily more accurate, but it is certainly harder for outsiders to verify. VAR sits squarely in that zone. It is transparent visually but not transparent in its interpretive process.

In Southeast Asia this is even more visible, because resources differ between matches and between rounds. A team can be denied a penalty in a round without enough camera coverage and then be awarded a similar penalty in a round with a full VAR room. From outside it looks like referee inconsistency. From inside it is infrastructure inconsistency.

Esports: regulation running behind the money

There is a field I have followed for years but rarely write about, because its data is far harder to verify than football's: esports.

What concerns me is not the games themselves but the speed at which the betting market has grown relative to the speed at which governance has been built. A traditional sport takes decades to assemble a supervisory framework: athlete licensing, monitoring of unusual betting patterns, cooperation with investigators, complaint procedures. Esports moved from community tournaments to a global betting market in a far shorter span.

That gap creates a distinctive risk surface. A nineteen-year-old competitor can receive an approach from a third party through a private channel, in a tournament whose organiser has no integrity monitoring unit at all. Meanwhile the audience watches through streaming platforms with near-instant data.

In data terms, esports enjoys a huge advantage: every action is recorded frame by frame. In institutional terms, it is far weaker than traditional sport. The distance between data capability and governance capability is where risk lives.

I have no definitive conclusion here. I keep one observation: when market money moves faster than the rulebook, data can prove an anomaly but cannot by itself stop it. "Football and esports share a bloodline: the rhythm of a match never lies." But the rhythm only lies when nobody is willing to look at it.

The transfer window: when noise drowns the signal

We are currently in a transfer window, the period when this profession is tested hardest in the year.

The problem is not a lack of information. The problem is too much information of wildly differing quality, presented with identical confidence. A rumour from a social media account and a brief from a licensed agent can appear side by side, same format, same length, same share count.

My working method is simple. I do not rank rumours by how exciting they are. I rank them by verifiable evidence: whether a release clause has been triggered, whether the payment structure is lump sum or instalments, how much contract remains, how many clients an agent has at the same club, and how much wage-budget headroom exists.

When I write transfer analysis, I open with contract structure rather than a player's name. A transfer can be read correctly or incorrectly depending on whether a club is paying for the present or for three years from now. That is the difference tables cannot express, and the reason so much transfer analysis sounds reasonable while carrying no predictive value.

What readers need from a data person in this period is not a re-sorted rumour list. They need a filter. And the most honest filter I can offer is to state clearly what I know, what I do not know, and which additional facts would change my assessment.

The counter-intuitive corner: blank is not failure

This is the part I want to give the most room to, because it runs against my professional instinct.

For years I believed a good analysis was one with a clear conclusion. My temperament likes order, categories, a table arranged neatly. A clear conclusion gives security to writer and reader alike.

But that fourteen-page blank report showed something the industry rarely admits: the greatest value of an analytical system is not its capacity to produce answers, but its capacity to detect when an answer is fake.

Our profession generates a vast amount of analysis. Every tournament produces hundreds of reports, thousands of articles, tens of thousands of forum posts. Most carry a conclusion. A meaningful share of those conclusions rest on data insufficient to support them.

The good news is that in recent years I have seen a small but real shift. More young analysts in Vietnam and Indonesia are stating their sources, their sample sizes, their limits. They still face pressure to predict, but at least they know where they stand. That is a more important step than any algorithmic upgrade.

Perhaps I am too optimistic. But I think it is a grounded optimism.

What I carry with me

Back to Surabaya. I resent the fourteen-page report to the coaching staff with the blank cells untouched, plus a short cover note: the retrieval system failed, we lack the basis to analyse this match, and here are the minimum data points required before any assessment can be offered.

The head coach called back forty minutes later. He did not complain about the missing data. He asked one question: if he had to track a single metric in the next match, which one should it be.

I told him I did not yet have enough information to answer. And that was probably the most correct answer I gave all week.

I believe in models, but I pray before every match — because football is not an equation. And after that empty report, I began to believe that the most honest part of sports data work lies not in what we infer from numbers. It lies in admitting when the numbers have expired.

If readers following a tournament in this transfer window wonder why the same source leads to three different conclusions in three places, the answer usually is not the analyst's competence. It is which blank cells each person chose to fill in.