The Empty Spreadsheet in Marseille: Discipline in Football Data Work

In January, a transfer dossier reached my desk at two in the morning. The xG column was blank. The minutes-played column was blank. The fee column carried a number nobody had verified. The sender, an agent in Lyon, wrote exactly one line: “Give me a conclusion, we meet in the morning.”

I did not give a conclusion. I sent back a list of six missing data fields, plus one line: without them, every judgement is guesswork dressed in terminology. That night I sat looking at an empty spreadsheet and remembered the first time I understood that the silence of data is itself a signal.

In the summer of 2026, I learned to trust something nobody had named yet: xG. When Opta first published expected goals for Ligue 1, I did not believe it straight away. I hand-recorded 1,204 shots from 20 teams across the first half of the 2026-18 season, checked them against actual goals, and got a correlation of 0.84. Enough to build my own striker-valuation dataset. Colleagues said I reacted slowly. I need verification before use, and I have not yet found a reason to change that habit.

Football today runs on a data layer thicker than at any point in its history. Opta and Stats Perform supply xG, xGA, PPDA. Transfermarkt prices squads. Major leagues enforce PSR and FFP. Every transfer story can now carry ten numbers. But as the data layer thickens, expectations thicken with it: readers want a verdict, and writers are paid to deliver one. Nobody pays for a blank cell.

So I work from a fixed grid of questions. Nine layers, and each one opens only after the previous layer has data.

The tactical layer needs a squad, a formation, pressing intensity measured by PPDA, and chance quality measured by xG and xGA. Croatia won a tournament on low PPDA? Then PPDA is merely a single letter. In the 2026 World Cup semi-final against England, Croatia allowed 8.2 passes per defensive action while England allowed Croatia 12.5. I forecast a Croatian win built on extra-time pressing, and they won 2-1. I did not shout. I reopened the spreadsheet to hunt for outliers.

The financial layer needs broadcasting revenue, commercial revenue, the wage bill, net debt and contract amortisation. The results layer needs a league table and a form sequence over a minimum five-match sample. The landscape layer needs to know whether a club sits in the title race, the European places or the relegation group, plus squad value against direct rivals. The rules layer needs the applicable regime — UEFA’s FFP or the Premier League’s PSR — and the matching precedent, from the 115 charges aimed at Manchester City to the points deductions handed to Everton and Nottingham Forest. The dressing-room layer needs the owner, the sporting director, the head coach and the contract status of every senior player. The risk layer needs injuries, suspensions and fixture congestion. The media layer needs the source and the agent’s motive. The final layer — the transmission chain from academy through club to broadcasting and derivative markets — runs only when there is a concrete event to transmit.

The January dossier failed at layer one. No formation, no PPDA, no net debt, no coach’s name. A blank cell in a dataset is an unanswered question; a zero is an answer already given. The distance between those two things is the entire credibility of the person writing.

I am 66, old enough to know that a number never tells a story unless you ask it to. In 2026, when European football returned after the pandemic, I sat in Marseille analysing 81 matches played in empty stadiums during the 2026-20 season. Home teams won only 26% of them, against 43% before the lockdowns. An empty stand is the finest laboratory a data obsessive could ask for. My report reached a Ligue 2 club, and they used it to negotiate down the price of a young striker with a dazzling home record. The lesson was not that home advantage had vanished. It was that every statistical table from then on had to separate home and away numbers.

In 2026, when the trade press praised Achraf Hakimi for 142 sprints and 2.3 chances created per match, I dug into the data and found the corridor behind him empty for 34% of the time. Morocco stayed safe because their centre-backs ran above 31 km/h. The inverted full-back trend holds only as long as the back line is fast enough to cover. Against France, the opposition funnelled the ball straight down that right flank.

Three times in five years, I learned the same lesson: a correct conclusion does not rescue a broken method.

The counter-intuitive part sits here. When a report returns nothing but “insufficient information”, the aggregation layer behind it will very easily read that as “no risk”. Those two statements are worlds apart. Missing data means not yet measured; zero risk means measured and found clean. Confusing those two states is how a third-tier rumour becomes a certain transfer, and how, when the deal collapses, the blame lands on the player.

This industry also has a clear incentive to fill blank cells. An empty xG column sells no copy. A sentence reading “not enough evidence to conclude” generates no clicks. So the space gets filled with adjectives: class, killer instinct, hunger to win. Those words sound sporty and measure nothing. A player is a variable, the market is a function, but most of my life has been a constant.

This transfer window, try one cheap measurement. Count how many of the ten player dossiers you read state the sample size, the time range and the source of the number. I suspect the rate is lower than you think. That rate — I call it the blank rate — is the next-cycle signal I will be tracking. When the blank rate rises, the quality of transfer decisions falls, and no valuation model rescues that.