Football Data Audit: When Statistics Are Evidence, Not Decoration

In the technical area at a La Liga match in the 2026/25 season, the big screen flashed the familiar line of statistics: the home side had 63% possession and 18 shots. The stands nodded. In my notebook, only nine of those eighteen shots came from what I call real chances — inside the box, unchallenged, with time to set up the ball. The other nine were long-range efforts under pressure, the kind every probability model files under low value. One match, two ways of counting, two different stories, and not a single person in the stands asked where the numbers came from. I have worked in analysis for seventeen years, through eight World Cups, eight Olympic Games and countless league seasons. What I learned does not lie in which team ran more, but in who decides how to count. Modern football runs on a vast data pipeline: cameras and sensors record thousands of events per match, companies such as Opta and Stats Perform standardise them into shared definitions, and from there the data flows to broadcasters, to betting firms, to newsrooms, and finally to fans through a single share. Each link in that chain has its own motive, and motive does not always align with truth. Back in 2026, when I had just left the pitch to join Valencia CF's coaching staff as a tactical analysis assistant, a senior male reporter asked whether I truly understood Marcelino's pressing block or had simply come to decorate the room. I did not argue. That weekend I sent the staff a fourteen-page analysis showing the team lost 62% of possession down the left flank, and Marcelino reshaped his line-up for three matches in a row. No one questioned my gender again, but the bigger lesson lay elsewhere: I won not with feeling, but by rebuilding every passage of play into evidence that could be cross-checked. My job is to audit matches. Before asking why we lost, I ask what we prepared for. Before praising the player who ran the most, I ask whether the measuring system counted all the meaningless off-ball runs too. Data does not lie, but people who read data do. A coach can pick a flattering metric to defend a substitution; a newspaper can pick a shocking figure to lift clicks; a betting firm can pick a loose definition to widen the market. One match, and each side reads a different slice. Take xG, the metric that estimates the probability of a shot becoming a goal based on position and situation. xG is useful, yet it is also the most abused metric. A team can generate high xG without scoring, and instantly the conclusion appears that they were unlucky. But xG does not measure the quality of the finisher that day, does not measure psychological pressure, does not measure whether the opposing goalkeeper read the direction of the shot. Citing xG without naming the model, the version or the sample size is a polite way of lying. The most worrying part of football's digitisation lies in live data being sold to betting companies. It is the darkest side effect of the whole industry. A pass, a shot, a foul is recorded in an instant and piped straight to the in-play market, where money moves second by second. When the value of a measurement is set by the speed at which it is sold, the slower demand for verification becomes a luxury. People need the fast reading more than the correct one. In the transfer market the problem is even clearer. Every summer, thousands of rumours are pushed out at the speed of light, and most have no verifiable source. A transfer fee circulates with no one able to trace where it came from — an agent, a club, or an anonymous account. Meanwhile, the genuinely valuable deals tend to happen at small clubs, where no one is watching and no one is selling the data. The transfer race among the giants is largely a race of brands. I paid the price for the lesson about where data comes from. At the 2026 World Cup in Russia, for England against Tunisia in Volgograd, I predicted England would press high in Guardiola style. I forgot that the afternoon temperature reached 34°C. England's players ran an average of just 9.2 km, 1.8 km less than in their previous match, Tunisia produced five dangerous shots, and Harry Kane scored both England goals in a 2-1 win. Manager Gareth Southgate said afterwards that he had deliberately lowered the intensity because of the heat. I had analysed tactics on paper while ignoring the environment. Since then, every analysis of mine carries a section for non-tactical factors: weather, pitch, travel distance, congested calendars. In 2026, when the pandemic stopped Spanish football for three months, Valencia fell into a financial crisis and sale rumours flooded the press. Instead of panicking, I built a notebook analysing the last nine matches from pre-pandemic data, then compared it with the fitness picture when the league returned. When the 2026/21 season kicked off, my analysis that Valencia had to shift from a 4-4-2 to a 3-5-2 for lack of strikers was republished by a major football site. There was no miracle here, only inventory first and commentary after. But I must also confess a paradox of my own trade. We live in an age when an analysis can look complete in form yet be hollow in content. I have received documents dozens of pages long, full of tables, bolded headings and a tidy contents page, yet when I checked, every data field was empty and every judgement read insufficient information to assess. The danger is that such a report, if not flagged, can still flow downstream and create the impression that analysis has been done. A handsome shell is worse than an honest emptiness, because it makes people stop asking questions. That is why I always mark the level of certainty for each judgement. I do not write that a team presses poorly unless a PPDA figure repeats across matches — the metric measuring the passes an opponent is allowed before each defensive action, where a lower value means more aggressive pressing. I do not call a coach a genius for one lucky win; a system earns recognition only when it reproduces across matches and contexts. A rule written in blood, not in ink. And a conclusion deserves trust only when it survives a second and third test in another match. The counterintuitive view sits here: modern football is drowning in unverified data, presented so beautifully that readers forget to ask for the source. When every match carries hundreds of metrics, metrics stop being evidence; they become decoration. And decoration is accountable to nothing. The press room is no place for the timid; it is a place for those with numbers — but only when those numbers still keep their roots. I do not believe in something called analysis detached from its source. A table of numbers with no origin, no publication date and no clear definition is just a pretty picture on a wall — pleasant to look at, but holding up nothing. Football deserves more serious audits, where every measurement must answer three questions: who recorded it, when, and to what end. Next match, when you read a line of statistics, try asking for its source. If the person presenting it cannot answer, then the problem was never in the match. It is in the reader.