The Empty Report

2:14 a.m. in Marseille

The spreadsheet is open on my screen at 2:14 a.m.: column A empty, column B empty, the summary cell showing a zero. No player name, no competition, no date, no source. Nine pages of analysis, every page beginning with the same sentence: insufficient information to assess.

I am 66 years old. My trade is reading spreadsheets like this one.

People outside the industry imagine a transfer-market administrator spends his days calling, negotiating and closing deals. Most of my time is spent cross-checking data tables and finding the row that is wrong. Some nights the sheet is crammed with numbers and I still cannot write a conclusion. Tonight it was the opposite: the sheet was entirely empty, and I was obliged to reach a conclusion anyway. The conclusion was that there is nothing to say.

In my profession, that is the hardest sentence to write.

I am 66, old enough to know a number never tells a story unless you ask it to. Data only answers when someone asks the right question, and it declines very politely when the question is wrong. An empty sheet is not shouting something. It is waiting to be asked again.

That night I did not sleep. I reopened old notes. In the summer of 2026, when Opta first published xG tables for Ligue 1, I hand-recorded 1,204 shots from 20 teams over the first half of the 2026-18 season and compared them with actual goals. The correlation came out at 0.84. That was enough to build my own striker valuation model — and enough to teach me something else: had I recorded 1,203 shots, the figure would differ. Had I skipped one club, it would differ. Had I recorded in March instead of November, it would differ.

That is why I never trust a new metric before counting it myself.

The pipeline: an article passes through nine doors

Before going further, I should explain the machine I and a small group of colleagues run in Marseille. A football article passes through two stages. The first is deconstruction: the source article is reduced to structured fields — title, source, genre, one-line summary, author's stance, purpose, information points, entities named, time sensitivity, source quality. The second is deep analysis across nine dimensions: tactics and technique; club finance and the transfer market; results and the public-opinion cycle; league landscape and team positioning; rules and governance; management and the dressing room; risk profile; media narrative and expectations; and industry transmission.

Each dimension is a door, and each door opens only with a key. The key is a named entity, a number with a unit, a date specific enough to check backwards.

For the tactical door, I need a formation, an xG figure, an xGA figure, a PPDA value, a possession share, or one concrete in-game situation. Without them there is no object to analyse. For the financial door, I need a transfer fee, a contract structure, a wage bill, a wage-to-revenue ratio, net debt, or a club with public accounts. For the time door, I need a date — because a record without a timestamp never expires, sits in the system forever, and distorts period-over-period comparisons. A stale record without a timestamp can make a club overpay two million euros for a striker. For the source door, I need to know where the story came from: a reporter with direct agent access carries a different weight from a social-media aggregate. With no source, the record's evidential weight is not low. It is zero.

That night, none of the nine doors opened.

Door one: tactics, and the empty chair of a midfield

The tactical dimension answers four questions: is the system sophisticated, is it executed well, do the personnel fit, and what do the key numbers say. Four questions, four times the verdict read: insufficient information.

No formation. No pressing scheme. No in-game adjustment. No xG, no xGA, no PPDA, no possession. Not one quantitative metric.

The emptiness of the tactical dimension is not a conclusion about tactics. It is a conclusion about the input data. Those are entirely different things.

I have fallen into that trap before. In 2026, at 58, working the World Cup for a sports daily, I counted PPDA for all 64 matches. In the semi-final, Croatia allowed England only 8.2 passes per defensive action, while England allowed Croatia 12.5. I wrote that Croatia would win through extra-time pressing. They won 2-1. I did not celebrate; I reopened the spreadsheet to hunt for outliers.

Those outliers taught me a rule I have carried for eight years: Croatia won a tournament of low PPDA? Then PPDA is only one letter of the alphabet. A pressing metric is one letter, not the sentence.

But even to say something that modest, I need the metric in hand. That night I had nothing.

Door two: money, and the price of a name that does not exist

The financial dimension is where I have a professional edge, because I once sat on the other side of the negotiating table. My structure has four columns: broadcast revenue, commercial revenue, wage expenditure, net debt — each with a trend and a risk flag. That night all four read: insufficient information.

No club, no deal, no fee, no contract length, no wage tier, no release clause, no agent commission, no instalment structure, no sell-on percentage. For the ordinary reader this sounds obvious. For a professional it creates a dangerous space, because it is precisely the space where invented numbers get inserted.

I have watched this happen many times. A transfer rumour with no source, no fee, no term. Three days later it reappears with a 25-million-euro price tag. A week later it has a five-year contract and 90,000 euros a week. Two weeks later another outlet cites the number and credits "European media". The number now has a life of its own, and nobody remembers it was born from emptiness.

That is what I call false specificity — detail generated to fill a gap rather than to describe a fact.

In finance, the damage is not that a reader is fooled once. It is that the fake number becomes the benchmark for the next deal. A Ligue 2 club checks the market, sees a price established by "European media", and pays it. The benchmark did not exist. The money did.

One story here is mine. In 2026 I analysed 81 matches played in empty stadiums in the 2026-20 season. Home win rate fell to 26 percent, against 43 percent before the pandemic. I wrote the report. A Ligue 2 club, Le Havre, used it to negotiate down the fee for a young striker whose scoring record had been outstanding at home. They saved a meaningful sum. What stayed with me was a different question: if my report had been wrong, how much of that player's value would the market have taken from him?

An empty stadium is the finest laboratory a data obsessive can ask for. But a laboratory still produces wrong results if you forget it is only a laboratory.

Door three: results, and an opinion cycle with no material

The results dimension is the most over-claimed in football writing, because everyone who watched the match feels entitled to speak. That night the sample was zero. No match, no table, no form line, no fixture list, no opposition quality.

More importantly: a manager's sack-pressure index cannot be computed without three things — a manager's name, a club's name, and a date. None were present. Sack pressure is a structured concept: contract length remaining, severance cost, sporting director's standing, relationship with the board, league position against pre-season expectation. Strip out the structure and you are left with a feeling.

Then comes the part I consider most valuable: the divergence check between process and results. Is a team winning on chance quality or on luck? This is the single most predictive question in the whole analytical table, and also the most abused. The check is conceptually simple — compare actual points with expected points derived from xG and xGA. But I never say a lucky team will be relegated. I say it is drawing on an unstable source of profit. Between those two statements lies a large gap, and most bad analysis lives inside that gap.

That night I had neither leg of the comparison. The most abused dimension in football commentary stayed silent, which is the responsible outcome.

Door four: the league and the food chain

Professional football runs as a food chain: contenders, European places, mid-table, relegation. That night all four tiers read: insufficient information. No league was named. The only usable label was "football".

I usually compare three resources across clubs in one division: squad market value, financial power, and academy output. These decide a club's long-run position more than any short-term tactical factor. A mid-table club with a good academy faces two choices each summer: sell three young players and live off transfer profit, or hold them two more seasons to climb toward Europe and sell higher. The second has a better expected return but higher injury risk, so boards usually pick the first, because accountancy dislikes patience.

Players are variables, the market is a function, but most of my life has been a constant. My constant is the rule of not concluding without evidence.

Door five: rules, and a statute without a defendant

The most important rule systems today are UEFA's financial fair play at continental level and the Premier League's profit and sustainability rules domestically. Some precedents are well known enough that anyone in the trade remembers them — cited here as reference points, with context.

In February 2026 the Premier League charged Manchester City with 115 alleged breaches of financial rules spanning multiple seasons; the case went to an independent commission and remains unresolved. In November 2026 Everton were deducted 10 points for profit and sustainability breaches, reduced to 6 on appeal in February 2026. In March 2026 Nottingham Forest were deducted 4 points on similar grounds.

Those three markers share one trait worth remembering: sporting punishment is decided by the balance sheet, not by form. A team can win seven straight and still lose points for losses booked two seasons earlier.

To apply a rule you need an actor and a jurisdiction. Neither existed in that report, so the compliance risk level was not "low". It was undefined. And for professional purposes, undefined risk must be treated as elevated risk — a principle I learned from accountancy rather than football.

Door six: the dressing room, where data never looks

I saved this dimension for near the end because it is the one I believe in most and the market prices lowest. Transfer valuation models rate youth potential highly — age, minutes, goals, assists, sprint speed, distance covered. They rate almost nothing a GPS vest cannot see: dressing-room chemistry.

A club can buy the right player and still fail, because he does not speak the goalkeeper's language, eats alone, and is treated by three senior players as an interloper. Another club can buy a statistically ordinary player and succeed, because he is the piece a group has been waiting for.

In the empty report, not even one person was named. That detail matters: in a football article there is almost always at least one human name. Its absence suggests an extraction failure rather than a genuinely person-free source. I record that judgment at medium confidence — and I state the confidence level, because a judgment with a stated confidence level is worth more than a certainty without one.

Door seven: risk, and the risk that is not where you expect it

My risk matrix has six rows: sporting, financial, personnel, rules, public opinion, systemic. All six were empty. But the risk section is where the report becomes interesting, because it identifies a different category entirely: analytical-integrity risk.

Risk ranking requires at least one subject and one exposure. Neither existed. Marking this "low risk" would be a category error: an unanalysable input is not a low-risk input, it is an unknown-risk input, and unknown risk must be handled as high risk in operations.

Five warnings followed. First, downstream misinterpretation: a well-formatted empty report can be ingested as "assessed and cleared". Quarantine the record. Second, silent data poisoning: with no timestamp the record never expires and skews comparisons. Attach an ingestion time and a mandatory TTL. Third, source recovery: with title and source blank, re-extraction must start from the raw fetch layer. Fourth, false specificity: the temptation to complete the template with plausible detail — a 4-3-3, a 30-million fee, an 88th-minute header. Enforce the rule that every conclusion cites an information point. Fifth, schema drift: emitting blanks rather than omitting fields suggests the template renders unconditionally. Make the extraction layer emit an explicit status flag.

Of those three states — failed extraction, empty source, success — only the third belongs on a front page.

False specificity: the illness nobody diagnoses

False specificity is not lying. A liar knows he is wrong. The person who generates false specificity believes he is making the story complete, which is exactly why it spreads.

The mechanism is simple. Given a ten-field template with one field empty, the writer feels the blank is his failure rather than the data's. I understand the feeling, because I have been there. In September 2026 I wrote a player profile with data covering March to July but missing the last two months because the league was suspended. I did not write those two months. I wrote one sentence: the final two months lack reliable data. The editor called to ask if I was sure. I was. It may have been the most honest profile I wrote that year.

By contrast, I once saw a detailed shot-by-shot xG table published for a Southeast Asian match. The problem: that competition had no event-data provider. The table could not exist. Nobody checked. It was shared thousands of times because it looked professional.

A professional format is not proof of professionalism. A beautiful xG table can be false. An empty report can be true.

My practical advice: ask three questions of any football analysis. Where does this number come from? How many matches is the sample? Does the writer state a confidence level?

The only four-star rating

Four value categories were unratable: sporting value, industry value, timeliness value, reference value. Rating them even one star would imply some residual content worth evaluating. One category scored four stars: process value.

I agree, and here is why an empty report has high process value. It localises the fault: the domain label succeeded while every entity field failed, so the defect is in entity and information-point extraction, not classification. It tests the null-handling path: the pipeline produced no conclusions, which is correct behaviour. It flags a recognisable pattern: a labelled but entity-free record usually means a failed or truncated fetch — a cheap fix. And it demonstrates something few organisations possess: the courage to publish a zero.

I have worked with analytics teams where submitting an empty report counts as personal failure. In such teams nobody submits empty reports, and every field gets filled — including fields with no data behind them. Such teams look more productive. They are simply fabricating systematically.

Some matches are won on the pitch and lost in the data — I choose the data. And some reports lose on presentation and win on truth. I choose the second kind.

The summer of 2026, and the value of waiting

In the summer of 2026 I learned to trust something nobody had named yet: xG. When Opta first published expected-goals tables for Ligue 1, I was 57 and working in the transfer market in Marseille. Younger colleagues were excited. I was not in a hurry. It took me nearly four months to hand-record 1,204 shots and compare them with outcomes. A correlation of 0.84 was enough to build a striker valuation model based on chance quality rather than raw goals. Colleagues said I reacted slowly. I think the opposite: in a business where information moves faster than the ball, four months spent verifying a metric is the highest-yielding investment I ever made.

What I learned was not the 0.84. It was the feeling of waiting. Once you have spent four months with an incomplete table and refused to fill a single row carelessly, you never fill a row carelessly again.

It is also why, at 60, analysing 81 empty-stadium matches in 2026-20, I separated home and away metrics in every table and stated the limits of my report: 81 matches, one league, one exceptional period. I did not say home advantage had died. I said that without crowds it fell sharply.

Vietnamese readers and the flag in the stand

I was born in Vietnam and have worked in France for more than forty years. That duality gives me a useful vantage point. Vietnamese fans follow European football differently from a season-ticket holder in Marseille: you follow through a flag. You choose a club, a player, a story, and you stay with it for seasons. That has an advantage European analysts undervalue — patience. It also has a weakness, and I say this with respect as a man who has stood on both sides: you are quicker to trust a number because it appeared in a professionally formatted article.

In a major-tournament season the effect is stronger. A tournament's emotional cycle runs through four phases: emergence, acceleration, peak, backlash. A transfer rumour passes through all four in about ten days. A manager, in about three weeks. A young player with two group-stage goals, in four days.

Learn to read sources in tiers. Tier one: direct access reporters with club beats and two-sided confirmation — high weight, rarely sensational. Tier two: verified aggregators and newsrooms with at least one editing layer. Tier three: anonymous accounts, fan communities, authorless aggregation pages — not bad, but carrying no evidential weight. Tier three is where stories begin, never where truth ends. And an authorless transfer rumour is not a weak rumour. It is an empty string.

Door eight: media heat against factual base

The most useful ratio in my media dimension is social heat divided by factual base. If chatter is high and the factual base thin, the story is about to reverse. If the factual base is thick and chatter low, it is about to be noticed. I have tested this across many seasons.

That night there was no story to measure, so the source tier was undefined and the weight zero. People like to say "this rumour has a 20 percent chance of being true". I do not use that construction, because it implies a base from which to compute 20 percent. With no source there is no base, and the correct figure is not 20 percent — it is undefined.

Door nine: industry transmission

Industry transmission runs in three layers: upstream academies and talent supply; midstream clubs and competitions; downstream broadcasting, commercial and derivative markets. It requires a triggering event. Without one there is no chain, and the correct output is to say no chain was drawn rather than to draw a plausible one. A plausible transmission chain is the content most easily misread as a forecast. And a note I attach to everything I write: this is description, not a betting recommendation. That report contained no odds data whatsoever.

What I found at the bottom of an empty table

First, a question about my own trade. If an analyst is paid to produce conclusions and produces none, has he worked? I believe he has, and that the work deserves payment even when the output is hard to see, because it stops a chain of false specifics before it is born. In medicine that is diagnosing correctly that a patient has no disease rather than prescribing to be reassuring.

Second, an observation about clubs. Over twenty years they have become better at collecting data and worse at saying "we do not know". The cause is structural: when financial reporting goes to a board every quarter, an empty cell looks like a loss. So it gets filled. And the filled cell returns eighteen months later as a bad signing.

Third, an observation about myself. At 66 I still read spreadsheets at two in the morning. At this age people want their stories tidy. I find I prefer the ones that cannot yet be finished.

Signals for the next cycle

First: the rate of records with a domain label but no entities. This is the pipeline's health metric. If it rises, the problem is not one article but the fetch layer.

Second: the arrival of an extraction status flag. When the upstream layer distinguishes failed, empty source and success, every downstream conclusion becomes more trustworthy.

Third: the share of analyses that state a confidence level. This is a cultural indicator, not a technical one, and it predicts the quality of a sports press better than any traffic number.

Fourth: the gap between popularity and factual base in transfer stories. The hottest deals are usually the thinnest.

And one signal I keep for myself: the number of empty reports I issue next season. The higher it is, the more comfortable I am about the system I run — paradoxically, because each empty report is one false detail that was never born, one fake xG table never shared, one club that did not overpay two million euros for a price that did not exist.

A cancelled match is not lost points; it is a lost page of a diary. But a blank page is not a lost page. It is a page we chose to leave blank, knowing that writing nonsense on it would ruin the whole notebook afterwards.

If you have read this far and feel short-changed because there is no club, no player, no fee — I confess I considered adding a few details to make it easier to read. I considered it for about two minutes.

Then I remembered I am 66 and remembered the 1,204 shots I once counted by hand. I typed this part instead of that one.

Tomorrow, another spreadsheet. Column A will have numbers, or it will not. My job is to know when to count, and when to pour a glass of water and wait.