When the Data Sheet Is Blank: A Lesson on the Limits of Sports Analysis

On Tuesday evening, I opened a thirteen-page deep analysis file that a colleague had sent over. The framework was complete: tactics and technique, player form, tournament structure, the world landscape, rules and institutions, the coaching staff, the risk surface, the media narrative, the industry transmission chain. Every table was in place, rows and columns as straight as a cadastral map drawn with a ruler. But in every cell, where a figure should have sat, there was the same cold line of text: N/A – insufficient information. Thirteen pages. Not a single fact. I sat still in front of the screen for a long while. What rose first in me was not discomfort, but an old professional itch. Analysts like me are trained to fill gaps. An empty cell is a challenge; a dash is a wound. Instinct pushed me to type in a probability, a percentage, a judgment — anything that looked certain enough to make tomorrow morning's meeting less awkward, to make the report look finished. Then I realized: this is the hardest test the profession can set. Not a test of how well I calculate, but a test of whether I dare leave a gap alone. A PROFESSION TAUGHT TO FILL GAPS In eighteen years in the trade, I went from a green data editor in Shanghai to a data consultant for clubs. That road taught me a great deal about football, and even more about the people who do this work. Modern sports analysis runs on an almost religious belief: everything can be measured. People measure distance covered, touches, goal probability, even players' heart rates and sleep. Every week, hundreds of new metrics are born, named, and sold to broadcasters and clubs like luxury goods. I have sat through three-hour seminars on machine-learning models and walked out still unable to say which team would win at the weekend. In Vietnam, the wave arrived later but no lighter. Post-match reports from the V.League are now crowded with figures: passes, pass-completion rates, duels won. The national team is dissected metric by metric after every SEA Games or AFF Cup. Fans have grown used to reading a match through colourful cells, and gradually forget that most of those cells tell only half the story. The problem lies here: once you are used to everything being measurable, you start to believe that what cannot be measured is not worth trusting. And at some point, the data person is pushed into choosing between two bad options: admit there is nothing to say, or invent a figure that sounds plausible. That thirteen-page file was the embodiment of the second option, but in blocked form. My colleague did not invent. She left every cell empty, and by doing so, she sent me a message stronger than any table of numbers. Vietnamese fans, I have understood after years of writing for them, do not lack passion. They lack time. They need a piece they can read in ten minutes that still lingers the next day. A surprising metric is a better opener than a long preamble. But that metric must be real, sourced, and able to survive scrutiny. That is why I never write an analysis just to fill space before a deadline. THE SHANGHAI LESSON: LOOK AT STRUCTURE, NOT THE SCORELINE In 2026, I was twenty-five, a data editor for a new football outlet in Shanghai. I was eager, confident, and not yet afraid of anything. That July, my club — then managed by a Portuguese strategist named André Villas-Boas — crushed an opponent 4–0. I wrote the same night, praising that pressing football as a masterpiece. My lines were full of energy: the home side pressing, the home side winning the ball, the home side scoring like slicing bamboo. I cited the scoreline as the most damning proof. I ignored one metric. It is called PPDA — the number of passes an opponent is allowed before each defensive intervention. That day my team posted 8.2 PPDA, a figure that looked good. But I never asked the extra question: our opponent defended deep, so they deliberately conceded the ball, and our pressure was never truly tested. Three days later, my team lost 1–2 to the bottom club. Not through bad luck. Because the pressing system could not sustain its rhythm, and the defensive structure behind it exposed cracks that a 4–0 win had covered up. My boss called me in and said something I remember to this day: You looked at the scoreline and not at the structure. Shanghai 2026 is not a scar; it is a map redrawing how I see numbers. From that day I built myself a hard checklist: every tactical piece must carry at least three advanced metrics, and the result must never be the only evidence. The summer of 2026 was the most expensive tuition I ever paid to learn: clean data cannot save a dirty hypothesis. That checklist, many colleagues say, made me rigid. They are right. But it is also the only thing that kept me from turning a 4–0 win into a wrong lesson. RUSSIA AND THE xG TRAP In 2026, I was twenty-six, sent to Russia as a data reporter for the World Cup finals. It was the first time I stood inside a major tournament with all the tools in hand, and the first time my own data betrayed me. The quarter-final between hosts Russia and Croatia, on the seventh of July. Before kick-off, my model produced an xG of 2.4 – 1.1 in Croatia's favour. I wrote a confident prediction that Croatia would win in normal time. The match ended 2–2 after a hundred and twenty minutes, and Russia fell on penalties. I did not sleep that night. I rewatched fourteen knockout matches, taking notes minute by minute. And I found something that chilled me: nine of those fourteen matches produced outcomes completely different from the model's expectation, once I added two variables the model ignored — the fitness cost of thirty extra minutes of extra time, and the invisible pressure of playing at home. xG measures chance quality. It cannot measure how tired the legs are in the hundred-and-tenth minute, nor the roar of sixty thousand fans pushing a defender half a step back. Russia taught me that the variable is not in the spreadsheet; it is in the player's pulse. I wrote the piece The xG Trap: When Numbers Lie, and more than twenty football pages shared it. But what I kept for myself was not the fame; it was a new habit: every article must include a small section pointing out the limits of the data I am using. THE PRESSING GHOST OF THE PANDEMIC SEASON In 2026, when global football froze, I was twenty-eight, heading a data content team at a sports platform. For six months I rewatched over a hundred old matches, just to hear something fans normally do not hear. In the empty-stadium matches at the start of the 2026/21 Bundesliga season, tracking data showed teams pressing twelve per cent higher, but the effectiveness of that pressing fell eight per cent. The reason was deeply human: without the roar, players lost a source of psychological energy, and their surges forward became mechanical rather than instinctive. When the stands were empty, I heard the sound of pressing footsteps most clearly under the pandemic night. Those footsteps rang out louder than ever, but they stepped to the wrong rhythm. I wrote the series Football in the Time of COVID: No One Hears You Shout, the Data Still Listens. My boss tasked me with building an automated writing system for closed-door matches. I refused. I proposed a model combining tracking data with remote coaching-staff interviews, because I believed some things only insiders can explain. That proposal earned me a promotion, but it also made me rigid: I forced every article to include a data-collection methodology section, a rule my colleagues found tiresome. Perhaps I overcorrected in the opposite direction. But the lesson stands: strip away the noise before saying anything meaningful. MERINO IN THE 119TH MINUTE AND THE LIMITS OF THE MODEL In 2026, I was thirty-two, a senior data expert at a Shanghai sports-analytics firm, advising several broadcasters. In the Euro 2026 quarter-final between Spain and Germany, our model rated the situation leading to the 119th-minute goal at a probability of 1.2 per cent. Mikel Merino headed into the net, from a Dani Olmo cross. A situation the algorithm treated as almost harmless. I wrote the piece that same night, asking whether artificial intelligence could ever predict a goal born from a seemingly harmless phase. I took Spain's successful header data from their last fifty matches and showed that nineteen per cent of those goals came from positions the model classified as impossible to score from. The piece was shared by many European analysts and opened a collaboration with a Spanish data company. But what I learned was not about reputation. I learned that macro data can never replace a player's tactical instinct in a split second, and that a probability model needs to be challenged by people, not canonised by them. Since then I have moved to a big-data, small-story structure: open with a counter-intuitive metric, go deep into the mechanism that produced the situation, and end on the human decision pressure of a fraction of a second. BADMINTON AND THE LESSON OF THE SILENCE BETWEEN RALLIES I write about badminton for the Chinese market, where audiences are used to metrics like shuttle speed, rally length, and unforced-error rates. Once I analysed a match featuring Nguyễn Thùy Linh — the pride of Vietnamese badminton — and noticed something interesting: most decisive points did not come from the best rallies, but from the silences between rallies. It is the time when a player breathes, wipes sweat, walks back to the service line that actually decides the match. Badminton data gave me exactly the lesson football data gives: what matters often lies in the gap between events. A fully blank analysis sheet, then, is not necessarily useless. It is the silence between two rallies, waiting for someone patient enough to listen. THREE METRICS AND ONE QUESTION My checklist has three tiers. The first is outcome metrics: goals, points, win rate. The second is process metrics: PPDA, xG, passes into dangerous areas. The third is context metrics: strong or weak opponent, home or away, days of rest. The most common beginner's mistake is stopping at tier one. The subtler mistake is stopping at tier two and forgetting tier three. And the fatal mistake is merging all three into a single metric and calling it truth. On top of that, I always ask one question before writing: in what circumstances was this metric measured? Fitness, refereeing, tournament pressure — each can change the meaning of a number. The same pass-completion rate tells a completely different story in a match where the opponent pushes high versus one where the opponent sits deep. Vietnam won the AFF Cup in 2026 and 2026, and each time I saw metric tables sprout like mushrooms. Some used pass-completion rates to prove a possession style. Others used goal counts to prove attacking power. Rarely did anyone place the two side by side and ask why they told two different stories. EFFECTIVE RUNNING DISTANCE In the empty-stadium matches, I began using a concept I named myself: effective running distance. Total distance only says a player moved a lot. Effective running distance says he moved to the right place at the right time. A midfielder who runs twelve kilometres but mostly chases the ball is entirely different from one who runs ten kilometres but is always at the hot spot half a second before his opponent. Raw data cannot tell the two apart. The eye that can read a match can. Two columns: hypothesis and evidence If the four stories above taught me anything, it is this: every mistake of mine stemmed from forcing data into a pre-set conclusion. My personality — the type who likes to control process, to keep everything in a frame — makes me especially prone to this trap. I want the story to close neatly, so I bend the numbers that lean the wrong way to fit the frame. The solution I adopted is to split the report into two clear columns: one for the hypothesis, one for the evidence. When the data deviates from the hypothesis, I keep the deviation and name it a blank point. I do not round it, trim it, or hide it in an appendix. That is why, when the thirteen-page file arrived full of empty cells, I was not flustered. I recognized it as honest. And in this profession, honesty is a kind of data more precious than any number. Emotion is a data column There was a period when I believed emotion was noise. A data person who lets emotion in easily loses objectivity — I used to think so. But the Russian night taught me the opposite: the human state is not something to be removed, but a layer of data equal to any other. What I do now is encode emotion into behavioural indicators. A player's body language after conceding — shoulders dropping, steps shortening — is a signal that can be recorded. The rate of technical decline in the final fifteen minutes is another signal. Repeated errors in the same area of the pitch are a third. When I place them side by side, they stop being subjective commentary and become a readable data column. And this is where I settle my belief: data can measure a player's running rhythm, but only an eye that knows how to listen can measure his breathing. WHEN CORRELATION IMPERSONATES CAUSATION Here I must say plainly something my industry rarely admits: most of the relationships we present as laws are merely correlations, and correlation is not causation. A team that presses high and wins many matches does not mean high pressing creates victory. It may be that taking an early lead is what lets that team dare to press high. Data does not speak causation on its own; it speaks coincidence. This trap appears everywhere. In the transfer market, people praise the signing fee for a free agent as a bargain, forgetting that such a fee sits outside the core oversight of financial fair play. In sports business, representation contracts make athletes reluctant to voice their true opinions, and gradually a politically-correct marketing replaces personality. None of this appears in any metric table, yet it shapes the match and the whole industry behind it. The biggest blind spot of my profession is not a lack of data. The biggest blind spot is believing that enough data is enough to understand. A system does not collapse in one night; it cracks from the moment I stop questioning the foundation. SIGNALS FOR THE NEXT ROUND That thirteen-page file from Tuesday evening will be filled in. I will send my colleague a request for additional sources, and within days the N/A cells will turn into numbers. But I know I will keep a copy of the blank version, to remind myself that sometimes the most honest thing a data person can offer is a silence. In the next round, the signal I will track does not sit in the points column. It sits here: which team dares to leave a blank cell in its report, instead of filling it with a plausible-sounding metric. Because in football, as in my profession, the most dangerous thing is not not knowing, but not knowing that you are inventing.