Nine Dimensions of Football Analysis and the Lesson of an Empty Data Table

I opened the file at six in the morning, Marseille time, before the streetlights had finished switching off. The table had every column header in place: club, player, transfer fee, matches played, expected goals per ninety minutes, distance covered. Twenty-six rows. Every data cell was empty.

In twenty-nine years of watching professional football, that was the first time I had seen a statistical table complete in form and absolutely hollow in content. The structure survived. The truth did not.

The document described itself as a football-domain analysis. The only populated field was a single word: football. No headline. No source. No timestamp. Not one club or player named. Nine analytical dimensions had been built like nine drawers, and all nine were empty.

I sat still for a while. Then I understood that what I was holding was not an unfinished analysis but a system failure set in a handsome typeface. A report with no data still reads very smoothly. It is simply wrong.

That is the real subject of this piece.

Context: the transfer window and the price of a confident tone

This month my inbox in Marseille is full of transfer bulletins. A few dozen headlines a day. Player X agrees to join club Y. Club Z prepares to pay one hundred million. Player W's agent is in London.

In the trade of transfer market data, I learned something that sounds paradoxical: the larger the volume of information, the lower the average quality. Not because people lie more. Because when everyone has to speak, those with nothing to say still speak.

The problem with an empty analysis is that it keeps the exact shape of an analysis. It has tables. It has sections. It has conclusions. It has confidence notes. If I cut the opening and sent only the tables to an editor on deadline, there is a good chance it would be published.

I have been on the other side of this situation. In 2026 I joined the sports department of Belgrade Television and sat next to a veteran editor who always asked one question before any story: "Who says?" Not "what was said", but "who says it". By 2026, when I had spent nine years hosting and producing "Football Night", I was still asking that same question of thousands of guests. It has since become the first rule of my transfer-rumour filter: information without a speaking subject is not information, it is an echo.

In October 2026 I published an analysis of Marseille against PSG on my personal blog. PSG won 3-0, while my expected-goals data showed Marseille generated the more dangerous chances: 1.94 against 1.21. Hundreds of comments poured in mocking the conclusion. Three months later PSG's underlying numbers fell and they lost 1-2 to Lyon. My read was confirmed, but I never treated that as a personal victory. It was the first time I understood that the hardest part of this trade is not finding the truth but holding on to it when nobody around you wants to hear it.

PSG won that year, but I chose to believe in the shots that did not go in.

Summer 2026 confirmed one more layer. A sports newspaper brought me in as a data expert for the World Cup. I tracked all three of Croatia's group-stage matches and recorded that they had run 318 kilometres in total, the highest in the tournament, while their average speed in second halves dropped seven percent against first halves. I warned that they would collapse in extra time. Croatia reached the final, and in that last match against France they ran eleven kilometres less than their opponents and lost 2-4.

Croatia 2026 taught me that heroes have biological limits too.

Both stories carry the same lesson: a correct conclusion is only trustworthy when the route to it was recorded. Without the route, a conclusion is just a guess wearing makeup.

Nine dimensions, and the minimum conditions that bring each one to life

The document I opened that morning had nine analytical dimensions pre-built. As a design, it is a good framework. I use almost the same skeleton daily in Marseille, renaming the parts to suit the French market.

The nine are: tactics and technique; club finance and the transfer market; results and the public-opinion cycle; league landscape and team positioning; rules and governance compliance; management and the dressing room; risk profile; media narrative and expectation; and finally transmission across the football industry.

The point I want to press is elsewhere: every dimension has a minimum data threshold. Below that threshold, a dimension is not "weak", it does not yet exist. This is the checklist I run before writing a single line:

• Tactics and technique needs at least one named club or player, a match or training context, and ideally one measurable metric such as expected goals, pressing intensity, or possession share.
• Finance and transfers needs a club name, a player name, a fee or wage figure, and a contract length. Miss one of those four and every market-value comparison becomes meaningless.
• Results and opinion needs a competition, a current standing, five to ten recent results, and the date of the source.
• League landscape needs a named league and at least two clubs playing in it. A single club standing alone is not a landscape.
• Rules and governance needs a specific incident: a transaction, a financial allegation, a sanction, plus the name of the governing body involved.
• Management and dressing room needs a named coach or sporting director, plus player names if contracts or injuries are in play.
• Risk profile needs at least one of the six dimensions above to be populated. You cannot rate the risk of something unidentified.
• Media narrative needs a headline, an author's stance, and a source tier.
• Industry transmission needs one discrete event with named parties: a transfer, a sanction, an appointment, a commercial deal.

The table I opened that morning had all nine rows. All nine failed the very first threshold.

A professional vocabulary, translated into human speech

One reason that empty analysis looked credible is that it spoke the industry's language fluently. I keep a personal rule: every time I use a foreign term, I must translate it into one Vietnamese sentence a skimming reader can grasp. If I cannot translate it, I have not understood it.

Here are the terms that appeared in that document, with my translations:

• Expected goals (xG): an estimate of the probability that a shot becomes a goal, used to measure chance quality separately from finishing ability.
• Expected assists (xA): the expected-goals value of the chances a player creates for teammates.
• Expected goals against (xGA): the defensive counterpart to xG, measuring the quality of chances a team allows.
• PPDA: passes allowed per defensive action. The lower the number, the more aggressively a team presses.
• Financial Fair Play (FFP): UEFA's rulebook limiting losses and requiring clubs to break even.
• Profit and Sustainability Rules (PSR): the Premier League's equivalent regime.
• Transfer amortisation: the accounting practice of spreading a transfer fee evenly across the contract's years rather than booking it at once.
• Sell-on clause: a former club's right to a share of a future transfer fee.
• Buy-back clause: a selling club's right to re-sign a player at a pre-agreed price.
• Release clause: a fixed sum that, if paid in full, automatically triggers a transfer.
• Contract year: the final season of a deal, often tied to form swings or renewal brinkmanship.
• Third-party ownership (TPO): a non-club party holding economic rights in a player; banned by FIFA.
• FIFA virus: the pattern of players returning from international duty injured or exhausted.
• Tapping-up: contacting a contracted player without his club's permission.
• Multi-club ownership: one ownership group controlling several clubs, raising eligibility-conflict risks in shared competitions.

Fifteen terms. Not one of them is needed to state that an empty data file is empty. That document used all fifteen to decorate emptiness. It is the oldest technique in the trade: when you have nothing to say, say it in difficult words.

Numbers have no bias. Bias lives in the person who lacks numbers.

A decision tree: which dimensions to read for which kind of story

The most useful part of that document, and the part I kept, is a routing table. It argues that depending on the type of article, the priority order among the nine dimensions changes. I have tested this in daily work and found it holds.

• For a match report or tactical review, read tactics, results, media first. Finance and rules are almost irrelevant.
• For a transfer or contract-renewal story, read finance, league landscape, media, industry transmission. Tactics and results come later.
• For a financial-results or financial-fair-play story, read finance, rules, risk profile.
• For a managerial change, read management, results, media, risk.
• For a league-structure or governance story, read rules, landscape, industry transmission.

Why does this matter during a transfer window? Because most summer rumours fall into the second group, and the second group is the easiest to fabricate. A transfer rumour does not need to be right about tactics or results. It only needs to sound plausible about contract structure and cash. That is why the best-sounding rumours always arrive with a number attached: seventy million, eighty-five million, a one-hundred-million release clause.

My experience following both matches and the market in Ligue 1 shows a fairly stable pattern: when a deal is announced with the full contract structure, the story is usually right on the numbers and wrong on the timing. When a deal is announced only as a headline fee, the story is usually right on the idea and wrong on everything else.

The transfer market does not buy players. It buys stories.

Three principles for a null return

Since then I have set myself three principles, and I would recommend them to anyone doing sports data aggregation.

First, state the extent of the emptiness in the opening line rather than burying it at the end. Readers are entitled to know whether they are reading a sourced document or an unsourced one before they invest their time.

Second, never fill a blank cell with reasoning shaped like data. A sentence such as "this club is probably facing financial trouble" is not analysis. It is an assumption wearing analysis as a costume.

Third, convert the emptiness into a to-do list. If transfer-fee data is missing, write down where to look, who publishes it, and when that information normally appears. A to-do list is more useful than a wrong conclusion.

The risk matrix, and one risk that is not in the matrix

That document listed six risk families: sporting, financial, personnel, rules, public opinion, systemic. All six were marked "not assessable". Technically, that is the correct answer. A risk table with no event to measure cannot produce a risk level.

But there is a seventh risk that the document itself uncovered, and I think it is its most valuable contribution: fabrication risk. When a system is asked to "analyse" an empty input, it can still produce a document that looks highly professional: club names, transfer fees, expected-goals figures, sanction precedents. All of it generated by the system itself. And the danger is that the final reader has no way to tell a generated detail from a sourced one.

A report with no grounding is more dangerous than no report at all, because its format transmits false authority.

This is why I open every piece with a table of numbers. Not because I like tables, but because a table forces the writer to answer the first question: where did this data come from? If that cannot be answered, the table stays empty, and that emptiness indicts the writer faster than any confession.

Data is the only thing I trust after witnessing too many promises break.

The counter-intuitive angle: silence is not innocence

There is a misreading I want to flag, because I have seen it happen to colleagues.

When a risk table reads "not assessable" on all six rows, the eye tends to scan quickly and register: no risk. That is the gravest interpretation error in this trade. Not assessable means not assessed. It does not mean checked and found clean.

I once saw a parallel case while working with fitness data. A club did not publish its injury data. On many tracking sheets the cell sat blank, and blank was read as "no injuries". Three weeks later three key players were missing for a decisive match. The data was not wrong. The reader of the data was.

But I also have to argue against myself, because this is the inherent weakness of anyone who works in data.

My method has a clear limit: it only works when there is a table. A coach changes system at half-time because he saw something in a player's eyes, and that never appears in any metrics table I have ever built. A player misses a decisive penalty for private reasons nobody knows, and my numbers record the miss but not the cause. And an empty data file, like the one that morning, paralyses my entire method.

In October 2026, when hundreds of comments claimed expected goals were a con, I spent three months answering with a data frame covering twenty-three Ligue 1 matches. I won that argument. But I asked myself: if that data frame had also been empty, what would I have answered with? Nothing. I would have had to stay silent, and in this trade silence is counted as defeat.

That is our real paradox. Sports media rewards confidence, not accuracy. A person who is confidently wrong gets quoted more often than a person who correctly says "I do not know yet". And so, every transfer window, we pour more noise into a system that already cannot separate signal from static.

People see a comeback. I see a chart that is breaking.

The noise variable: what a data table can never measure

If the table is empty, what should I do? The honest answer is: do not write the analysis. In practice I still have to write, because it is my job. So I work in three steps.

First, I state the extent of the emptiness inside the piece and do not pretend to have grounding. Second, I shift from analysis to questions: instead of asserting what will happen, I list what needs checking and in what order. Third, I set a review date for myself: in ten days, which data should have appeared, and if it has not, which of my hypotheses is eliminated.

A reader once asked whether that is evasion. My answer: evasion is writing a conclusion you know has no grounding. Stating the limits of your data is the only honest form left.

There is one more variable my tables never capture, and I want to say it plainly here. The transfer window does not run on probabilities. It runs on relationships. A well-timed call from a respected agent can change a deal that every valuation model had ruled out. Dressing-room chemistry, a star accepting a pay cut to stay, a young talent turning down a bigger club because he believes in a coach, none of that lives in any column.

A risk model has never saved anyone, but it gives them a chance. And a chance is the best a table of numbers can offer a human being.

Signals for the next tracking cycle

I still keep that empty file in my working folder, named "lesson-from-july". It holds no information about football. But it holds something important about my trade: structure can be built in advance, truth cannot.

My next tracking cycle turns on four specific signals, and I list them here so anyone who wants to verify them can check for themselves.

First, the share of transfer reports that name all four elements: selling club, buying club, fee structure, contract length. I expect that share to sit below twenty percent at peak.

Second, the number of deals announced as a headline fee that later have to be revised downward. This is the single best indicator of market exaggeration.

Third, the gap between the moment a rumour appears and the moment it is confirmed or denied. The shorter the gap, the more reliable the source.

Fourth, and the signal I care about most: the number of analytical pieces that state their own data limits. If that number rises, our trade is improving. If it falls, we are moving faster towards a market where every number looks beautiful and none of them is right.

I am not predicting which club wins the title. I am not predicting which player moves where. Those belong to a zone that data can touch but never hold. What I can do is keep my own tables honest, even when a blank cell is the only correct answer.

That morning in Marseille, after closing the file, I wrote nothing. I called a friend who analyses football in Lyon and asked the exact question the 2026 editor taught me: "Who says?" She laughed and replied: "Nobody yet."

It took us ten minutes to confirm that we were both looking at the same kind of empty data. Then we moved on. Some days, the most accurate conclusion an analyst can offer is that there is nothing to say today, and that needs to be written down clearly.