The Data Void: When Football Mistakes Silence for Safety

On the night of 19 May 2026, at the Allianz Arena, Bayern Munich took more than twenty shots and generated an expected-goals figure somewhere around 3.0. Chelsea left with the first Champions League trophy in their history. I sat in front of a screen in Shenzhen, rewinding the match three times, hunting for a single metric that could explain what had just happened. None of them could. Petr Cech saved Arjen Robben's penalty in extra time, Didier Drogba headed an equaliser in the 88th minute, and the shootout ended in favour of the away side. The post-match dashboard was packed with numbers and yet hollow exactly where it mattered: it could not measure the psychological collapse of a team that had just missed a penalty on its own pitch. That night taught me something the football analytics industry still refuses to learn properly fourteen years later. Bad data can be fixed. Missing data is the dangerous kind, because human beings are wired to read silence as safety. Football analytics has travelled a long road since that Munich final. Expected goals is now everyday language on every broadcast. PPDA measures pressing intensity, the number of passes an opponent is allowed before each defensive action. Player valuation models, injury forecasting and season simulations run behind the scenes at almost every leading European club. Each Premier League club maintains a data department of dozens of staff, with budgets that can reach several million pounds a season. And yet I still see pre-match previews stuffed with beautiful dashboards in which every metric carries a value and not one cell is blank. That is precisely when I start to distrust them. When a model has no empty space, it is usually because someone filled it with an assumption rather than with data. I forge my opinions on the anvil of data, swinging a blunt hammer, so let me be blunt: the biggest risk in modern football is not wrong data. It is that analytical systems only report what they measure and stay silent about everything else. A centre-back with no standout numbers gets undervalued, even though the model has never once measured his ability to organise a back line. A club with no financial red flag gets treated as healthy, even though its books were never loaded into the system. A few years ago I helped test a sports data platform. The night before launch, every data feed returned empty. No headline, no event, no entity. The interface still displayed all its fields, every placeholder was labelled, and the export button still glowed. Had anyone pressed it, the system would have produced a complete document with full section headings and conclusions, missing exactly one thing: the truth. I remember the chill of realising that an empty system can still speak fluently. THE TRAP OF EMPTY CELLS There is a principle in logic that football commentary violates every single day: the absence of evidence is not evidence of absence. Not finding a breach does not mean there is no breach. No warning indicator does not mean no risk. The Premier League brought 115 charges of financial rule breaches against Manchester City in February 2026, after a multi-year investigation. Before that moment, in the eyes of most supporters, the club was a model of success. On the balance sheet, everything checked out. The problem is that proving a number is legitimate requires knowing exactly where it came from. When the origin is obscured, the data still looks clean, but it loses verifiability. Everton is the mirror case, and a far more painful one in sporting terms. On 17 November 2026 the club was docked 10 points for breaching profit and sustainability rules. On 26 February 2026, after appeal, the deduction was cut to 6. On 8 April 2026, Everton received a further 2-point deduction for the following financial period. Eight points stripped in a single season, more than enough to reshape a relegation battle entirely. Nottingham Forest were docked 4 points on 18 March 2026. The number is smaller; the meaning is not. A newly promoted club that had spent to survive was immediately dragged backwards by the financial system. In Italy, Juventus hit the regulatory wall twice. In 2026, in the Calciopoli affair, the club was stripped of two Serie A titles, relegated to Serie B and began the season with a points deduction, initially 30 and reduced to 9 on appeal. Seventeen years later, a capital-gains and salary-manipulation case saw Juventus docked 15 points on 20 January 2026, the ruling annulled on 20 April, then reimposed at 10 points on 22 May of the same year. Around the same period, the club was excluded from European competition. What all of these cases share is that they only surfaced when somebody bothered to dig into the data nobody wanted published. The analytics laptop on the touchline detected nothing. The only red flag appeared in the accounts department. THE BURIED xG PULSE Expected goals is a beautiful tool, and also a textbook victim of the empty-cell trap. When a team scores far above its xG for ten straight matches, we praise its character. When it does so for a season, we praise its philosophy. When it does so for three seasons, we praise its legacy. But somewhere in the data, an anomalous conversion rate is always waiting to regress to the mean. Goalkeeper performance is the cleanest illustration. A keeper who saves more than his expected goals conceded can sustain that form for one season, sometimes two. But expected goals conceded does not measure reflexes, reading of the game, or composure in a shootout. It measures shot placement and match context. The rest is empty space, and that empty space decided Bayern against Chelsea in 2026. I am not a prophet. I only see three steps ahead of the dance of chaos. Across the last three matches of a team in resurgence, their PPDA has dropped noticeably, meaning they now allow fewer passes before each defensive engagement. That signals a midfield pushing higher, accepting risk to win the ball early. If their expected goals conceded also ticks up, the model sounds the alarm. But the model does not know the side just lost its first-choice centre-back to a hamstring injury, information that lives in no column of any spreadsheet. That is why I always read a dashboard twice. The first time to see what it says. The second time to see what it hides. Every number is a match waiting for someone who knows how to listen, but only if that person understands the match also contains an audio track that was never recorded. WHERE I COULD BE WRONG Let me argue against myself before anyone else does. There is a perfectly reasonable counterpoint: if an analytical system cannot measure something, then failing to raise an alarm is correct behaviour, because science should only draw conclusions within the scope of its data. In that reading, the data department is blameless. The fault lies with the reader, who silently assumed that a complete report equals a complete reality. But operations do not work that way. Data departments have now pushed deep into the dressing room, into transfer decisions, into the morning tactical meetings. Their conclusions are often detached from the real rhythm of a team, because that rhythm is built from variables that cannot be digitised: internal friction, fear of losing a place, family pressure, the ego of a star just benched. I could be wrong in another way too. It is possible the points deductions in England and Italy are merely the visible tip of a system slowly becoming more transparent, and that within a decade the data sources will be thick enough to fill nearly every empty cell. If that happens, my argument becomes obsolete, and I will happily admit it. I have been wrong in public before, and I know how to survive it: go back to the data, find where I misread, and write again. But right now the evidence leans my way. Three of the clubs hit hardest by points deductions in Europe over the past two decades share one trait: their analytics departments worked very well. WHAT I EXPECT NEXT I expect at least one more Premier League club to be docked points over the 2026 to 2026 financial period within the next two seasons, and at least one further European case involving multi-club ownership structures. This is a falsifiable prediction, not a curse. In parallel, I expect expected goals to evolve into a metric with mandatory confidence intervals, and analyses that refuse to publish their own uncertainty to lose their audiences. Supporters are getting more demanding, and they have started to notice that a model silent about its own blind spots is an incomplete model. When the whole world looks in one direction, I open the door they never thought to knock on. This time the door sits behind an empty cell on a spreadsheet, where nobody bothers to ask a question. If you are reading a report where every cell is filled, ask yourself which one was filled with belief.