The Empty Spreadsheet: What Remains When Football Has Nothing Left to Measure

The Empty Spreadsheet: What Remains When Football Has Nothing Left to Measure

Eleven in the morning, Madrid time. I open a spreadsheet on my second monitor and find fourteen data fields, all carrying the same content: N/A. No team names. No player names. No metrics. No opinions. An analysis was sent to me for dissection, and inside it there was nothing but white space.

I stared at it for about three minutes. Then I did what any data analyst does when handed an empty dataset: I started asking why it was empty. Who built it, for what purpose, and what had happened that left fourteen fields without a single character of information.

The obvious answer is that the process failed. But in football, the obvious answer is rarely the right one.

On July 1, 2026, I sat in front of a television in Madrid and watched Spain hold 75 percent of the ball, complete more than a thousand passes, take roughly twenty shots, equalise through an own goal by Sergey Ignashevich, and then go out on penalties against host nation Russia at Luzhniki. I was seventeen, old enough to bet a friend that Spain would win 3-0, and young enough to spend that night reopening the match statistics and searching for something I had never heard of: expected goals.

The number that appeared silenced me. Spain generated about 0.7 xG from all those shots. Under 0.04 xG per attempt. Which means nearly every one of those shots came from outside the box, from tight angles, from situations where Russia's low block was already set and waiting.

A spreadsheet packed with numbers can still be empty of information. That was the first lesson. I once believed in absolute numbers, until a World Cup taught me that emotion is a variable too.

These two events, seven years apart, say the same thing. The value of data lies not in how many cells are filled, but in whether we have the courage to ask the right question. Data does not hand you answers, it only surfaces the question you were brave enough to ask.

The Anatomy of a Blank

Professional football analysis runs on three layers of data, and those three layers are usually collapsed into one in the public mind.

The first layer is raw data captured by optical tracking systems or in-ball sensors. Every second, every player leaves behind dozens of coordinate points. The second layer is event data, labelled by providers such as Stats Perform or Opta, with human quality control attached. The third layer is public data, the part fans and journalists actually see on statistics sites.

The distance between the first layer and the third is far wider than most people imagine. A single match can generate millions of data points, but only a few dozen numbers are ever published. The rest sits with clubs and providers under exclusive contracts.

Which means when you look at a statistics page and see a handful of metrics, you are looking at an extremely thin slice of the truth, curated by a chain of decisions you do not control.

And above all three layers, there is always a fourth: data that does not exist. Things never measured. Things nobody bothered to record because nobody had yet worked out how to price them.

N/A Is Not Zero

This is the most common interpretive error I encounter in my work.

Zero is a measurement. When a team registers zero shots on target, that is the result of a real observation: somebody sat and counted, and the count came back empty. Zero exists in the space of events.

N/A is the absence of measurement. It does not exist in the space of events; it exists in the space of process. When a cell reads N/A, I cannot know whether the team achieved anything. I only know that nobody recorded it.

The distinction sounds academic, but its consequences are concrete. When working with football data, I always separate three kinds of absence.

The first is absence through non-collection. The league has no tracking system, so positional data does not exist. This kind is neutral in intent.

The second is absence through non-disclosure. The club has the data but chooses not to share it. This kind carries clear intent.

The third is absence through non-existence. Nobody on earth has ever measured it. This kind is an opportunity.

Each of these three demands a completely different response, and collapsing them into a single N/A is the fastest route to a wrong conclusion.

Three Matches, Three Blanks

I want to walk through three cases I have pursued, because each represents a different kind of blank, and each taught me something the spreadsheet could not say.

The first is Spain against Russia in 2026. This was a blank of the third kind: at the time, xG was not yet familiar to the Vietnamese public, nor even to most Spanish fans. The metric existed in the event-data layer, but it did not exist in the conversational layer.

That gap produced a specific psychological consequence. Without xG, people were forced to explain defeat through something else. And the easiest thing to reach for is a story about character, about luck, about a curse. That is why, after that match, Spanish media spent weeks discussing the national team's mentality rather than asking why twenty shots had produced 0.7 expected goals.

Fans look at the scoreline; I look at the probability. After 2026, I know both can collapse.

The second case is Italy at the Euro played in 2026. I was twenty-one, finishing a thesis in sports statistics. I calculated Italy's PPDA and got an average of 7.8, the lowest in the tournament. Simply put, opponents were allowed fewer than eight passes before Italy attempted to win the ball back.

The interesting part is that Italy were not the team that ran the most. They were not the team with the highest total distance covered. They were the team that ran in the right places.

I wrote a five-thousand-word piece for my personal blog predicting Italy would win because the pressing line was so synchronised. A sports journalist in Madrid shared it, the piece pulled twelve thousand reads in forty-eight hours, and a Spanish football site paid me one hundred and fifty euros to republish it. It was the first time I understood that my data had a market price.

But there is a detail in that story I rarely tell. Italy's low PPDA was not only because they pressed well. It was low because opponents chose to circulate the ball deep in their own half, and Italy deliberately let them. The most beautiful number in my article was actually the product of a defensive choice, not an attacking one.

Italy did not win Euro 2026 through luck; they won because they turned data into a way of playing.

The third case is the crowdless season of 2026 and 2026. I was twenty, working remotely as an intern for a small sports data firm in Madrid. My assignment was to compare Real Madrid's home performance before and after crowds returned.

The result kept me in front of the screen for a long time. With empty stadiums, Real Madrid averaged 1.9 goals per home match. With crowds back, that number fell to 1.3. Meanwhile, expected goals barely moved.

Read only the spreadsheet and you conclude that crowds make a team score fewer goals. But stable xG means the quality of chances created did not change. What changed was conversion.

I presented the finding in an internal meeting. My manager praised it. A colleague pushed back, arguing the sample was too small and the crowdless season too anomalous to support any conclusion.

He was right, statistically. I responded by extending the sample to ten La Liga seasons. But I also had to concede something I have since written at the bottom of every article: widening a sample does not make a wrong hypothesis right. It only shrinks the error, making a wrong hypothesis harder to detect.

In 2026, with empty stadiums, football exposed systems and choices.

The Trap of Filling Blanks With Stories

The human instinct is to fill gaps. Cognitive psychology calls it the need for closure, and football is a perfect environment for it because football is full of gaps.

When a team wins three matches in a row with the same shape, nobody knows whether the shape or the fixture list is responsible. But the coverage will talk about the shape.

When a player scores four goals in five games, nobody knows whether that is genuine improvement or regression to the mean. But the coverage will talk about improvement.

Words like character, spirit, winning DNA, class, and fire are not descriptively wrong. Those things are real, and I have changed my position enough to accept them as part of the structure rather than noise. The problem is that they are routinely used as filler for the cells we are too lazy to measure.

Emotion is a variable. But a variable you cannot measure remains a variable you cannot measure, however real it is.

The second trap is more dangerous: confirmation bias when hunting for the counterintuitive angle. When you pride yourself on seeing what others miss, you start selecting evidence to protect that reputation. I have made this mistake. I went through a phase where every piece needed a surprising conclusion, and I kept bending data into narrow angles to manufacture surprise.

The fix I apply to myself is simple but uncomfortable: write the conclusion first, then actively hunt for data that could break it. If nothing contradicts it, the conclusion may still hold. If something does, I rewrite.

Home Ground and the Data Nobody Records

There is one variable that modern tracking systems still cannot fully quantify, despite its clear effect on results.

That variable is the crowd.

In event data, a home match and a neutral-venue match are recorded with the same set of metrics. Player positions, passes, shots, all structurally identical. But the pressure that twelve thousand people behind the goal exert on the legs of a twenty-two-year-old has no column in the spreadsheet.

In practice, I have to admit that home advantage is something I measure indirectly rather than directly. I measure it through the gap between xG and actual goals. I measure it through late yellow cards. I measure it through whether a team plays shorter or longer when leading.

Every one of those measurements is an estimate from the outside. They surface questions; they do not deliver certainties.

That is why I envy basketball analysts. In basketball, home and away shooting splits can be compared directly because shot volume is larger and pace is more stable. In football, a match contains only twenty to thirty shots from both sides combined. That is far too small a sample to separate signal from noise.

A team is not a collection of metrics; it is a system breathing through every pass. Basketball is the natural habitat for data analysis, while football is the natural habitat for ambiguity.

N/A as a Signal

This is the part I consider most important, and the part it took me years to dare to state.

An empty stretch of data is not always a problem to be fixed. Sometimes it is the information itself.

When a club publishes its own xG but not per-player xG, that is a choice. When a league publishes financial fair play standings but not detailed wage structures, that is a choice. When a federation does not release individual injury data, that too is a choice.

The shape of the blank is itself data. It tells you who has an incentive to hide what, and what is considered important enough to keep sealed.

I once wrote a piece on La Liga showing that top clubs publish a great deal of attacking data and very little individual defensive data. The reason is obvious: attacking data sells shirts, defensive data helps opponents. That blank has a name, an owner, and a commercial purpose.

In Vietnamese football, the blank is far larger but different in nature. Most of it is non-collection rather than concealment. Domestic competitions lack a public xG system with enough history and stability for time-series analysis. The metrics I use for La Liga can only be hand-calculated in personal spreadsheets for a handful of matches.

But here is the paradox: precisely because Vietnamese data is hard to obtain, analysts in Vietnam often have to watch football more carefully. They have to remember, observe, write things down. In Spain, where every metric sits on a website, the tendency is to read numbers instead of watching the match.

Easy access to data breeds laziness in observation. That is something professionals in Spain rarely say out loud, but I have seen it in myself.

These two lenses complement each other in ways neither side recognises. Vietnamese observers assume data is a luxury reserved for European football. Spanish observers assume reading data is instinct, forgetting that the instinct is gradually replacing the ability to see.

Both sides are half right, and both are wrong in the same place: they believe there is one correct way to understand football.

What I Carry Forward

I do not believe in turning football analysis into a contest between data and emotion, where whichever side wins holds the truth.

I believe in keeping both truths standing side by side. Data reveals patterns. Intuition reveals exceptions. A system built only on patterns will be broken by the first exception. A decision built only on intuition cannot distinguish a real trend from the illusion of the last three matches.

A title is built with data, but it is saved by intuition from thousands of hours of watching.

Back to the empty file from the start. Hand it to a machine and it returns an error. Hand it to someone who has worked the trade long enough and they will ask three questions: what kind of blank is this, who created it, and what becomes visible if I assume it was created precisely so that I could not see.

The season is at the stage where the table still says little, but process data has started talking. Teams generating good chance quality with poor results tend to rise. Teams posting good results on unusually high shooting conversion tend to fall. That is the pattern anyone willing to look can see.

The harder part lies elsewhere. Which team is hiding a dressing-room problem behind a good run. Which player is fielded in the right position but the wrong role. Which manager is winning on individual quality while the system rots. None of that appears in any public dataset. It lives in the blanks that only a viewer who watches enough, and patiently enough, can sense.

I still keep a spreadsheet tracking match-by-match xG, started when I was seventeen. It has thousands of rows and hundreds of empty cells I have never filled. I used to treat those empty cells as debt. Now I treat them as a map of questions not yet asked.

In the coming weeks, as the season passes through its heaviest stretch of fixtures, some team will drop points at home against a weaker opponent, and a headline will explain that the team lacked character. I will open my spreadsheet and check the xG from that match. If xG was high and goals did not come, I will not write about character. I will write about how many empty weeks sit in this season's calendar, and whether anyone on the coaching staff is responsible for counting accumulated minutes per player.

And if xG was also low, I will ask a harder question: why a club with that level of investment cannot create chances on its own ground. That is a question data can open but cannot answer. Only eighty minutes of rewatching the tape can.

An empty spreadsheet is not the end of analysis. It is the beginning of observation.

---

Limitations

What I have set out above draws on personal observation across ten years of following football, together with hand-built spreadsheets I have maintained since 2026. Most of the cross-season comparisons rest on samples smaller than the standard required by academic statistical research.

The conclusions about home and neutral-venue conversion need further verification using full tracking data, which I do not have access to across all of La Liga. The observations on Vietnamese football are based on remote viewing through streaming platforms, not on direct contact with coaching staff or club data departments.

This article offers no recommendation related to betting. Football always carries probability, and probability is not certainty. Every conclusion about a team's future can be wrong.