When the Data Table Goes Blank: How Football Analysis Lies to Itself

When the Data Table Goes Blank: How Football Analysis Lies to Itself

2:47 a.m., March 12, 2026, in a fourth-floor flat in the Ruzafa district of Valencia. On the left screen, a Segunda División match on matchday 30. On the right screen, the live data dashboard I rent by the month: pass maps, expected goals, pressure charts, distance covered, refreshing every thirty seconds. At minute 63, every cell turned grey. No expected goals. No pass counts. No heat maps. Only the clock ticking steadily on, like the heartbeat of a machine already brain-dead.

I kept writing. Over the next 47 minutes I produced roughly 1,400 words about the home side's pressing shape, about the midfield being opened up in the inside channel, about a centre-back playing eight metres higher than usual. When the feed returned, I checked every sentence. Three claims matched the data. The rest was memory, prejudice and confidence.

That was the moment I understood something this trade rarely says out loud: a data gap is never silent. It always gets filled, and the person filling it is us.

Context: an industry built on the assumption that data is always there

European football has come a long way since Prozone arrived in the Premier League in the early 2000s, when clubs hired students to click a button for every pass. Today, a match in a top European league generates roughly three thousand manually coded events, plus millions of positional points from optical tracking systems. Stats Perform, StatsBomb, Hudl, SkillCorner and dozens of smaller providers sell these packages to clubs, broadcasters, bookmakers and to writers like me.

Based on my experience watching matches both from the stands at Mestalla and through a screen, I see the industry running as a four-stage pipeline: capture, extraction, modelling, and finally storytelling. Extraction is the most fragile stage. It depends on correctly calibrated cameras, on a connection that does not drop, on a file format that does not fail. When that stage breaks, the other three do not stop.

They do not stop because they cannot. A bulletin has to go out at 11 p.m. An analysis has to be published before readers move on to the next match. So when a cell is empty, people fill it with whatever is closest to hand: intuition, memories of similar games, and personal credibility.

I used to think this was a problem specific to freelancers with no verification. But after talking to three analysts working at optical data centres in Spain, the answers were the same: partial signal loss happens far more often than technical reports admit. And the most common fix is still interpolation.

An empty cell is never neutral

An empty data cell carries an invitation. It does not say "I don't know." It says "guess." Psychologically, humans tolerate ambiguity far less well than they tolerate a wrong judgement. We would rather be wrong than left hanging.

This is the mechanism I call confident interpolation. When numbers are missing, the brain does not leave the slot empty. It samples from the nearest available store of memory, usually matches with a similar context, similar pitch, similar opponent, then labels that sample with a feeling of certainty. That feeling is not proportional to the amount of evidence. It is proportional to how many times we have told a similar story.

There is an example I still use when teaching students in Valencia: give two analysts the same match, but let one watch with the data dashboard and the other without. They will produce two different descriptions of the same defensive line. The one with data will talk about the distances between units. The one without will talk about attitude and spirit. Both are partly right, but only one of them knows what he is missing.

Everyone sees the ball; I see the person holding the pen that draws the match. And when the numbers are gone, that person is drawing from his own memory.

Seven shots and four goals

On July 15, 2026, at Luzhniki, France beat Croatia 4-2 to win the World Cup. I filed my first piece at 3 a.m. Spanish time, arguing that France were champions but undeserving. The numbers I cited: France had the ball for roughly a third of the match, took seven shots and scored four goals.

Six years later I reread it, and the value is not in the conclusion. The value is that those seven shots said nothing about the match. It is a real number, quoted accurately, and it carries almost no information about how France created chances. France exploited the space behind Croatia's midfield with long diagonal passes exactly three times in the whole game, and all three led to a goal or a penalty.

No mainstream dashboard displayed that metric in 2026. No column was called quality of pass after ball recovery. So the writer, even the writer with data, had to choose between talking about what was measured and talking about what happened. Many chose the first, because the first comes with a nice graphic.

France won, but what I said in 2026 still stands. That team won the trophy with things that were not in the box score, and the fact that we argued about it for years mostly proves the box score failed to do its explanatory job.

Salzburg 2026 and a metric that did not yet exist

In 2026, as a second-year sociology student, I watched the UEFA Youth League final involving Salzburg. My flatmate only cared about the score. I was drawn to something else: at times, seven Austrian players sprinted forward together in the final thirty metres, none of them touched the ball, and the opponent lost possession after four passes.

I wrote a 2,000-word piece on my personal blog claiming Spanish football would have to borrow this idea within three years. It was mocked as looking too far ahead. Three years later, high pressing was routine in La Liga.

The point is not that I was right. The point is that in 2026 no metric measured what I was watching. PPDA, the measure of passes allowed per defensive action, only became widespread in Europe a few seasons later. Before the metric existed, all I had was my eyes and patience.

The paradox is this: without eyes, I would not have seen it. With eyes alone and no discipline in note-taking, I could not have retold it. Data does not create a judgement. It only confirms or refutes a judgement that already exists. And when the data does not yet exist, the writer carries the responsibility alone.

The tactical con from Spanish youth football is present all over Europe. But it was only recognised after someone dared to trust his eyes before a metric backed him up.

Valencia 2026-04 and what the dashboard leaves out

In March 2026, La Liga was suspended indefinitely. I could not write a line for three weeks. Then one night I decided to rewatch every Valencia match from the season they won La Liga under Rafa Benítez, to find what turned that side into such an enduring machine.

I took notes by hand for weeks, colouring starting positions, and found something no mainstream stats page displayed at the time: Benítez's rotation that season was so heavy that by today's numbers it would be read as instability. The same eleven almost never started two matches in a row.

Today's dashboards have a minutes column, an appearances column, an average position heat map. None of them show the logic behind rotation. They record the outcome of a decision, not the reason. Someone reading the numbers sees eleven different names each week. Someone watching the whole season sees an idea about workload and timing.

The distance between those two views is the distance between data and understanding. The numbers are not wrong. They are simply answering a different question from the one people assume they are answering.

VAR, fifty frames and sixteen centimetres

There is one area where the illusion of completeness is clearer than anywhere else: offside technology. From the 2026 World Cup in Qatar, FIFA introduced a semi-automated offside system with twelve dedicated cameras tracking twenty-nine body points, sampled fifty times per second. It sounds infallible.

But here is the arithmetic anyone can do. Fifty samples per second means the gap between frames is two hundredths of a second. A player sprinting at thirty kilometres per hour, about eight point three metres per second, moves close to seventeen centimetres in that interval. A third of a foot.

The system compensates by interpolating between frames, and that model works well. But it must be said plainly: the offside line the viewer sees is a reconstruction, with an error margin, presented in graphics that make the margin look like zero. The person who drew that image never appears on screen.

The problem is not the technology. The problem is how media uses it to end arguments. A line drawn by an algorithm, with a known tolerance, is treated like a court ruling. And the attacking instinct of players, which is nourished by accepting risk in the narrowest of margins, is being adjusted by a probability model they never get to see.

Distance covered and runs that lead nowhere

This is where I argue most with colleagues in Spain. Distance covered and sprint counts are packaged and sold to viewers as metrics of effort. In reality they measure movement, not effectiveness.

A full-back who runs twelve kilometres in a match may simply be a player who is repeatedly out of position and has to run back to fix it. A central midfielder who runs only nine kilometres but makes three runs at the right three moments may be the most important player on the pitch. The dashboard ranks the first above the second. The match does not.

I once sat next to an analyst from a La Liga club during an evening of football. He said something I wrote down immediately: if you pay for a metric, you will tend to defend it in the meeting. That effect does not only happen at clubs. It happens across this trade.

Empty stadiums in 2026 and the voice of tactics

The pandemic took away the roar, but it gave back creativity in despair. When the stadium has no crowd, I hear the clearest voice coming from tactics. Eleven players no longer had a window to show off in front of the stands. They only had a job.

In June 2026, when La Liga returned after nearly three months, I watched the first match and noted something unusual: long passes spiked, individual dribbles dropped sharply. But when I checked the stats afterwards, I realised I had described the cause wrongly. I thought the absence of a crowd created structural strictness. The data later showed the decisive factor was a schedule of one match every three days, with fitness not yet recovered after the long break.

That was the most expensive lesson I ever paid for myself: both hypotheses were plausible, and only data could separate them. If I had not bothered to check, I would have left behind a false claim that sounded wonderful, and it would have lived in my archive for years.

Where I could be wrong

There are three places where my argument is weakest.

The first is the assumption that judgement before data is always valuable. It is only valuable when paired with serious later checking. A writer with good instincts who never cross-checks builds an archive of randomly correct and systematically wrong claims.

The second is underrating interpolation. The models data providers use when signals drop were built by people better at this than I am. In most cases their models are more accurate than my memory. If I dismiss their value, I am repeating the very mistake I criticise: trusting a single source.

The third, and the one I think about most, is assuming the audience wants the truth. Football is entertainment. A bulletin containing only the line "insufficient data to conclude" can be perfectly accurate and simultaneously useless to someone who got up at six in the morning to rewatch a match. The demand for narrative is larger than the demand for verification. I may be proposing a standard the market will not pay for.

I do not write to be agreed with; I write to open a door someone else locked. But that door only means something if someone walks through it.

What I think happens next

Over the next two seasons, I expect a new role to appear in European clubs' analytics departments: a data integrity officer whose job is to publish which signals dropped and which metrics lack a sufficient sample to conclude anything. I also expect major data providers to start selling packages that include coverage and interpolation rates per match, starting with the betting market, where error has a price.

And I expect something smaller but verifiable: the most influential football analysis piece in Europe next season will be one that ends with the sentence, no conclusion can yet be drawn. If that happens, this trade has moved forward a step. If it does not, we will keep writing 1,400 words in the minutes when the dashboard is blank, and keep calling it analysis.