The Gap Between the Numbers: How Football Analytics Fills Empty Cells with Stories

At two in the morning on 8 March 2026, I sat in front of a spreadsheet with 110 rows and four blank columns. It was Bundesliga data from the 2026-21 season, played behind closed doors because of Covid-19. Research by Kai Fischer and Justus Haucap, published in 2026 and covering exactly 110 Bundesliga matches without spectators, showed that home advantage measured in points had fallen sharply. I wrote the 43% decline into my notebook.

But there was one column I could not fill: successful pressing actions per player. At the time, the league did not supply complete tracking data for every match.

I almost made it up.

Not made it up by doctoring a number. Made it up by writing a sentence that sounded perfectly reasonable: "This team has maintained pressing intensity at the same level as last season." That sentence flows, reads like analysis, and has nothing behind it to verify. If I had published it, nobody would have objected. That is the frightening part.

Football analytics has never been so confident. European clubs run data departments with dozens of staff. Broadcasters put xG on screen the way they put the scoreline on screen. Fans argue using PPDA without needing to know what it measures. The consensus is simple: more data means better decisions.

I do not dispute that consensus. I dispute what sits directly underneath it.

Bad data was never the biggest problem. The problem is empty data, and the way this industry handles emptiness. In a transfer meeting, a blank cell is not read as "unknown". It is read as "find a way to fill it". On a television segment, a blank cell is not read as "no basis". It is read as "we need a line". And so a lattice of reasonable-sounding sentences grows on top of an empty dataset.

None of this is new. The scale is. Fifteen years ago people said "from my observation". Now they say "the data shows". The second phrase sounds ten times more certain, while the underlying basis may be identical. I once predicted France would beat Argentina 4-3 in the 2026 World Cup round of 16, a match played on 30 June 2026 in Kazan, where Kylian Mbappé scored twice and won a penalty. The 900-word piece drew 120,000 views and launched my writing career. But the foundation under that prediction — 27 sprint bursts, an Argentina defence reacting 0.4 seconds too slowly — came from a clip sample I cut myself. I never disclosed the sample size.

This is where I want to be blunt: effort metrics are the most abused category of data in modern football, because they are always available even when real analysis is not.

Distance covered. Sprint count. Number of runs above 25 km/h. These numbers get packaged as an "effort index" and sold to audiences as a measure of quality. But ineffective running produces beautiful numbers. A midfielder who covers 12.4 km in a 0-3 defeat has a heat map splashed across the pitch — he ran a lot because his team did not have the ball, not because he was good. A full-back with 30 sprints may simply have been chasing balls he himself lost. The metric measures movement, not meaning. It is presented as measuring both.

The second problem is inconsistency between providers. In the same match, on the same shot, xG can differ between models because each provider defines a "clear chance" differently. Major providers do not fully disclose their model weights. Which means that when two experts argue using two xG values from different sources, they are arguing about two definitions of the same word, and neither of them knows it. The reader only sees one number. A number that looks extremely decisive.

The third problem is data geography. Major leagues have near-total tracking coverage. Leagues in Southeast Asia, V.League among them, still rely largely on manual tagging. An analyst working on a V.League match does not lack expertise — he lacks raw data. When he publishes, his output looks identical to the output of an analyst with a full dataset. Readers cannot tell the two apart. Neither could I, until I started asking each of them where the numbers came from. I read the data, and the data whispers a name nobody has chosen — but sometimes that name is only the name I wanted to hear.

And this is where the story turns a little more serious, financially. The data department is the first thing cut when budgets shrink, and it is cut quietly. Jiangsu FC won the Chinese Super League in 2026 and ceased operations in February 2026. Between those two markers, what disappeared was not just a club. An entire layer of data collection and processing staff was dissolved. Those who remained in the industry moved to working with old datasets, or with third-party datasets that nobody re-checks.

Nobody re-checks. That is the crux.

In finance, a missing data cell blocks a transaction until it is verified. In football, a missing data cell is usually handled with a sentence. That sentence goes into an article, from the article into a transfer decision, from the transfer decision into a club's balance sheet. I have tracked this chain long enough to know it is not hypothetical.

I could be wrong in three places, and all three are worth stating.

Controlled estimation of empty cells is sometimes better than leaving them empty. Statistical imputation methods are legitimate tools when they are declared transparently. If a club states plainly, "we imputed pressing data for seven missing matches", that is science. What I object to is imputation that happens in silence.

Complaining about data gaps may also be a privilege of wealthy leagues. A small club that cannot afford data still has to take the pitch, and can still win. Sometimes they win because they lack data: no model tells them how they must play. Sitting deep is not cowardice, it is how intelligent people wait for fools to charge — and intelligent people do not always need a chart to know what they are doing. Tactics are not a formula. They are the answer to a reverse question: what does the opponent fear most?

And I am the one who used a sample of 14 clips to describe Argentina's defence. If I applied my own standard to my 2026 piece, I would have to call it speculation with decoration. I still defend the conclusion. I do not defend the process. Every prediction can be wrong. Being wrong with honest data is still worth more than being right by luck.

In football, the most obvious thing is usually the least verified. People look at the league table; I look at the gap between the numbers — and the most notable gap over the next three years will not be in cells that contain numbers, but in empty cells filled with confident prose. My prediction, and it is checkable: within the next two transfer windows, at least one major deal will be publicly justified with a metric for which the provider does not cover the player's league at all. Anyone who wants to verify it should keep the press clippings.