When the Spreadsheet Goes Silent: The Fragile Line Between Data and Truth in Football

In October 2026, in a small apartment in Marseille, I sat in front of a screen with an open spreadsheet. The Marseille – PSG match had just ended 3-0 in favour of the visitors. But when I loaded the data into my model, everything stopped. The expected-goals column returned empty cells. Not zeros. Not negatives. Simply empty. A silent spreadsheet. And in that silence, I realised something more than a decade in the profession had never taught me: an empty dataset is not evidence of anything — it is a refusal. The truth is, many football readers live among such empty cells every day without knowing it. I am telling this story not to talk about one specific match. I am telling it because the feeling of looking at an empty spreadsheet has followed me for eight years, through every transfer analysis, every player evaluation, every late night spent comparing figures that nobody else wanted to cross-check. PSG won that year, but I chose to believe in the shots that did not go in. The context matters more than people realise. At the time, event-stream data was not as widespread in France as it is now. Large data providers divided the market by competition, each with a different definition of key metrics. Some matches returned full positional data, others only half, and a few — like Marseille – PSG that night — suffered a provider sync failure, turning every advanced metric field into an empty cell. Technically, it was a data-pipeline incident. Professionally, it was the moment I had to choose: publish an analysis built on incomplete data, or admit I did not have the basis to speak. I chose the second option. I wrote a short piece stating plainly that my dataset was incomplete, that what I had observed on screen was not enough to draw a firm conclusion, and that I would wait. Three days. Wait for the source to be patched. Wait until every empty cell was filled with a traceable number. Hundreds of comments poured in. People said I was cowardly, that I did not dare to conclude, that a woman working in data only knew how to stay silent when there was no ready-made answer. But I held my ground. Because I believe something very simple: if you conclude while the data is still empty, what you produce is not analysis — it is speculation dressed up in terminology. Three days later, the data arrived. And when I reloaded everything into the model, the picture that emerged differed sharply from what the 3-0 scoreline had told. Marseille created more dangerous chances than their opponents. PSG won through an abnormal conversion rate, not through dominance. I rebuilt the entire evaluation frame across 23 Ligue 1 matches in the same cycle and showed that the Paris club was winning heavily on a conversion rate far beyond any forecasting model. My conclusion at the time: this trend was unsustainable. Three months later, PSG's metrics dropped and they lost 1-2 to Lyon. My assessment was confirmed — not because I was a prophet, but because I had enough data to see what an empty spreadsheet could not say. Since then, I have set myself an inviolable rule: never analyse when the data lacks coverage. And from that, I realised the football world is full of silent spreadsheets nobody notices. There is a more dangerous kind of silence than a technical failure. It is the silence of metrics published halfway — enough to create an impression, but not enough to verify. You see a player score 15 goals this season. You do not see that nine came from the penalty spot, that he touches the ball 18 times per match, that his running distance in the second half of matches drops 12% versus the first. Those empty cells are not displayed. They are left outside the spreadsheet, and the reader only sees the highlighted part. My first-hand experience watching Ligue 1 matches in stadiums gives something different from reading reports: when you sit in the stands, you see the player slow down. You see him stand still for two seconds before accelerating. You see the breathing. The spreadsheet does not record those two seconds, but those two seconds are the story. Let us return to the 2026 World Cup, the event that put my name on the data-expert list of a European sports newspaper. I tracked all three of Croatia's group matches. The team ran a total of 318 km — the highest figure in the tournament at that point. But when I split the data by half, a pattern emerged: average speed in the second half fell 7% versus the first. Not a slight drop. A systematic one. Croatia still won, still scored, still made people believe in extraordinary mental strength. But the spreadsheet was whispering the opposite. Croatia 2026 taught me that heroes also have biological limits. I wrote a warning: if Croatia went deep, they would collapse in extra time. Not because they lacked courage, but because the human body cannot run 120 minutes repeatedly at peak intensity without paying a price. At first, nobody in the editorial team noticed my note. Then Croatia reached the quarter-final against Russia, played 120 minutes, and needed penalties. Then the semi-final, another 120 minutes. Then the final against France, where they ran 11 km less than their opponents and lost 2-4. After the tournament, an editor called me: "How did you see that?" I answered honestly: I did not see it with my eyes. I read it in a spreadsheet that many consider dry. Without it, I too would have been swept up in the heroic narrative like everyone else. This is the point where I want to pause, because I know it is easily misunderstood. Many people think that working with data means denying football's emotion. That is wrong. I love the game the way people love a symphony: I love the low notes that were never written down. The problem is not emotion, but that emotion is often used to fill the gaps that data leaves behind. When there is nothing to measure, people measure with legend. When there is nothing to verify, people verify with belief. That is when risk appears. In my work as a transfer-market administrator, this kind of risk appears with terrifying frequency. Every transfer window, hundreds of rumours rise, and most are built on empty spreadsheets. An account posts that a player is on his way to a big club. No named source. No specific timestamp. No contract structure stated. Just a claim, and a crowd chasing it. The transfer market does not buy players — it buys stories. This is true in both directions. A club is willing to pay 60 million euros for a 22-year-old striker not only because he scored 18 goals in a season, but because the story around him can sell shirts, sell tickets, and sell hope to a city that is hungry. That story is rarely tested by a model. It is tested by a feeling in the meeting room, where a president says: "I feel he is going to break out." That feeling is on no spreadsheet. I have witnessed dozens of deals worth tens of millions decided under such conditions. That is exactly why I built my own personalised risk scorecard for every player profile. It is not meant to deny a coaching staff's intuition. It is meant to turn intuition into probability — so that when wrong, people know where they were wrong, and when right, they know why. A risk model saves nobody, but it gives them a chance. My scorecard has four main variable groups. The first is the biological fitness profile: age, height, weight, history of tendon and muscle injuries, muscle mass gained over the past three years. The second is the tactical profile: actual minutes played in a top-tier league, touches in dangerous zones, ability to drop deep and join build-up. The third is the psychosocial profile — the hardest to measure but often decisive: language adaptability, off-pitch stability, relationship with the agent. The fourth is the destination-club structural profile: what formation that club plays, whether this player is a genuine priority or merely a contingency option. I often tell young colleagues in France that the fourth group is the most neglected. People analyse the player thoroughly, but rarely analyse the club the player is joining with the same seriousness. The result is contracts that are right at the individual level but wrong at the structural level — and such deals are usually dismissed as "not a good fit," when the real issue is that nobody checked the structural cell before signing. In modern football, there is a trend I watch with growing concern: inverted wingers are becoming so common that teams look alarmingly alike. I have analysed hundreds of goals to understand this, and what I found is not that inverted wingers are ineffective. On the contrary, they are highly effective in specific situations. The problem lies in a double consequence: when every flank inverts, the wide space empties, and opposing defences can compress the central lane more easily. A metric that does not show up in highlight reels, but shows up clearly in the event-stream data, is a significant rise in blocked through-balls. I believe traditional wingers are being wrongly erased. Not because they are outdated, but because current valuation models cannot measure their value. A winger who dribbles the full width, who creates space in areas nobody watches, may have a lower individual xG than an inverted winger. But if you measure the space he opens for teammates, the number changes. The problem is that most models have not yet measured that variable. Data is the only thing I trust after witnessing too many promises break. But I must admit this fairly: the data I trust is not empty data. It is data collected with a clear method, traceable to its source, and published with its gaps rather than hiding them. An honest model is not one without empty cells. An honest model is one brave enough to say: "This part, I do not yet know." This is the line that I think many in the industry cross without realising. They fear silence. They fear the blank spaces. They fear the moment of having to tell an audience: not enough data to conclude. So their path is usually: if data is missing, add rhetoric; if evidence is missing, add claims. That is how an analysis becomes propaganda, often without anyone intending it. For Vietnamese readers following European football, I think this matters more than daily transfer rumours. Because you are absorbing a huge volume of information from sources most of you cannot verify from Vietnam. When a rumour appears in English or French and is translated, it is often presented as fact. At the original level, it may be a two-sentence item with no source, written to drive engagement. So what should readers do? I do not advise you to doubt everything, because doubting everything is also a form of blindness. I advise you to learn to read the gaps: when an item names no source, that is a gap. When an item has no specific date, that is a gap. When a number is given without a unit or comparative context, that is another gap. You do not need to know the answer. You just need to recognise you are reading a spreadsheet with unfilled cells. And this is what I learned after eight years working with football data: most of the time, the truth is not in the published number, but in the question of who holds the unpublished numbers. Numbers have no bias. The bias lies with those who lack them. In this transfer window, I am tracking a few specific profiles I will not name. What is notable is not the names, but the structure of the deals being negotiated. I see more and more contracts built around variable clauses, performance add-ons, and conditional release clauses. They do not get flashy coverage like the transfer fee in the headline, but those details are precisely what determine a deal's real value. This is where I want to be clear about what I am observing. A contract with a low fixed fee but complex variable structure is usually presented in the media as a cheap deal. In reality, if the player performs, the buying club can pay far beyond market value. My valuation model weights that add-on portion heavily, because that is where real risk accumulates. One thing I always tell my students: never judge a transfer by the headline number. In daily work, I constantly have to choose between issuing an early judgment to gain a media edge, and waiting for more data to issue an accurate one. This industry rewards speed, not accuracy. The first to report is remembered. The one who reports correctly is overlooked. That is a professional reality I have never seen change. But I still choose slow. Because I have seen too many empty spreadsheets filled with promises, and seen the consequences when the season ends. There is a concept I borrowed from structural engineering that I learned while living in France: panic-proofing code. In high-rise buildings, systems are designed to work precisely at the moment everything collapses. The principle is simple: every bad situation is a variable, every decision is a line of code, and every system needs a code to protect itself when reliable input data runs out. Amid global panic, I choose to write code for safety. I apply that principle to football analysis. When the transfer market surges, when people race to break news, when a club suddenly spends 100 million euros on a player nobody mentioned six months earlier, my code triggers: stop, check the empty cells, verify the origin of every number. Not to deny the deal, but to understand what the deal really is before it becomes a story. In a transfer window shaped by new broadcast-rights cycles, every deal is no longer an isolated event. When rights packages are renewed, new money flows into the system, and new money always seeks the places most easily mispriced — usually young players unproven at the top level, and contracts negotiated under low-transparency conditions. I have watched this cycle at least three times in my career, and each time the bubble model is the same: value is pushed up on a narrated potential, and corrected downward when real data appears. What worries me is not the bubble. Bubbles are a natural part of markets. What worries me is individual decisions made on empty data, and those decisions often have lasting consequences for young players with no voice in their own moves. I once witnessed a case: a young player was transferred to a top European club at a fee considered reasonable. But when I checked his actual minutes data, I found most of his time had been in the reserves, or in matches already decided. The goal figure looked impressive, but the context behind it was not disclosed. Six months later, he had almost vanished from the first team. None of those who originally assessed the deal admitted they lacked data. They simply said the player did not adapt. That is the industry's sad truth: when an empty spreadsheet is filled with a hasty conclusion, the player bears the error. So what should be tracked in the next round of fixtures and the next transfer cycle? First, track running distance and pressing intensity in the second half of each match, especially during congested schedules. This is an early-warning metric the scoreline does not show. A team can win three matches in a row while its average second-half speed in the third match has dropped 8% versus the first. That signal does not appear in the table, but it foretells what is coming. Second, track the contract structures published in the final phase of the transfer window. Deals done in the last 72 hours typically have higher variable-fee ratios and weaker protective clauses for the buying club. That is a sign of decisions made under time pressure, not under data pressure. Third, track the correlation between chance-conversion efficiency and points won for clubs competing for European places. When a team has an abnormally high conversion rate but creates few chances, it is on a broken line, not a rising one. The world sees a comeback; I see a chart breaking. But I also admit something those of us in data rarely say: there are things spreadsheets can never capture. The moment a player looks into a teammate's eyes before a penalty. The moment a coach decides to substitute in the 88th minute while his assistant advises otherwise. The moment a team trailing suddenly changes rhythm with an action nobody planned. I cannot measure those with a model. I can only say that after each such moment, I trace the data again to see whether a pattern lay behind it. Sometimes there is. Sometimes there is not. That is why I keep one final principle in my risk scorecard: always reserve one empty cell for the unknown. Not out of laziness. Out of respect for the truth. A good model is not one that answers every question. A good model is one that knows exactly where it cannot answer, and says so without fearing a loss of credibility. In my next analysis, I will return to a more specific subject: the relationship between pressing intensity and muscle injury during congested schedules. I have enough data to speak on that. And when I have enough data, I will speak. When I do not, I will keep silent — until the empty cells in my spreadsheet are filled with traceable numbers, rather than retold stories. And if there is one thing I want readers to carry away from this piece, it is this: learn to look into the gaps in every piece of information you receive. The most important answer is often found where the spreadsheet is most silent.