Empty Data in Sports Analysis: When Numbers Fall Silent

October 2026, I was living in a university dormitory in Chicago, writing a football blog for myself, and had just spent an entire week obsessed with a match that probably no one in America cared enough to rewatch twice. Huddersfield Town beat Manchester United 1-0 at the John Smith's Stadium. The press told the story through a familiar template: a small club toppling a giant through luck. The visitors dominated possession at 78 percent, fired 19 shots, generated expected goals of 1.82, while the hosts managed only 0.35. But the win belonged to Huddersfield. The old tale of football's injustice was told once again. I downloaded the match data, rewatched the footage about twenty times, and found a number that appeared in no news report: Huddersfield completed 27 tackles in front of their own penalty area. Not 27 tackles in general, but 27 ball recoveries in the final thirty meters, precisely at the intersection between the defensive midfield line and the centre-back pairing, where every touch is paid for with a probability of error. That number was not a headline-worthy statistic, but it explained the entire match in a way xG never could. The journey to a final is not in the legs; it is in the distance they are willing to run. I wrote that line for the first time in the summer of 2026, and I still keep it today. The name Data Monk has been attached to me since that day. I set up a small website called "I Have a Number" and began writing about the metrics the mainstream media left behind. My first principle was born from that very match: when xG lies, every number must be interrogated from the beginning. Why an empty data table is scarier than a wrong one I was born in China and now live in Chicago, working as a data consultant for football clubs. My job in recent years has not been to put pretty numbers on a report, but to interrogate those numbers until they confess the truth behind them. In eleven years observing the sports analytics industry, I have learned one thing more important than any model: data is never in a hurry. It waits until you are sober enough to ask the right question. Last month I received an analytical document from the data department of a well-known esports organisation. The document had a tidy structure, a clear domain label, and an opening section written by an experienced editor. But when I opened the detailed content section to find the first data point, I discovered the information array was empty. No tournament name, no team name, no player name, no number, no date. The entire data section was blank, with only one surviving field: esports. My first reaction was to assume this was a single glitch. But when I checked seventeen other documents from the same source, I found a pattern: three of them showed the same condition to varying degrees. A repeating pattern is no longer a bug; it is a symptom. Modern sports analytics faces a paradox: the more automated it becomes, the more data is generated, and the fewer people can distinguish between "no data" and "data not processed." An empty table can result from a filter set too tight, a system failure, or simply a document that never existed in the first place. But to the final reader, the coach, the scout, the sporting director, that empty table can be misread as "nothing worth reporting," and that is when disaster begins. Dissecting an empty data table To understand why an empty table is dangerous, one must first understand the structure of data in modern sport. A standard football or esports analytical report is built on four layers of information: domain label, information point, entity, and source. These four layers are chain-dependent; each layer can only be processed if the previous one exists. The domain label is the only thing that remained in the document I received. It tells us the document belongs to esports. But the domain label is the least valuable of the four layers, because esports is a category containing dozens of titles that are mutually non-transferable in competition format, player metrics, and governance structure. League of Legends, Dota 2, CS2, Valorant, Honour of Kings, they cannot be analysed under a shared framework. Each title has its own metric system, its own patch cycle, its own analytical community. When I asked a colleague in charge of esports analysis at a New York organisation about the document, the answer made me realise the severity of the problem: "Ah, I thought the system had handled it." The "system" here is an automated pipeline with two stages. Stage one extracts events and information points. Stage two reads stage one's output to produce deep analysis. But when stage one returns an empty result, stage two keeps running, still outputs a report with a full structure, still attaches the necessary headings. The only difference is that there is no data inside. This is the crux I believe anyone working in this industry must etch into memory: empty data is not zero data. It is a signal that must be read correctly. When a metrics table holds no values at all, the first thing to do is not to fill the gaps with speculation, but to stop and ask why the gaps exist. In football, the same principle holds. When a player appears in no statistical table, no touches, no tackles, no chances created, the analyst's natural reflex is to conclude the player performed poorly. But in reality, this phenomenon has at least four possible causes: the player was injured early and left the pitch before data was captured, the player operated in a position the tracking system could not register, the player performed so well the opposition avoided his zone entirely, or simply the data was corrupted in transmission. These four scenarios lead to four completely different conclusions. And the poor analyst will choose the first, the easiest and most dangerous. The Sofyan Amrabat case and the lesson on data value To demonstrate the severity of the problem, look back at the case I experienced with Sofyan Amrabat after the 2026 World Cup. The tournament in Qatar saw a superb performance from Morocco, the first side in history to reach the semi-finals. At the centre of that success was Amrabat, who recorded 24 ball recoveries across five matches, a gargantuan figure for a defensive midfielder at a World Cup. In early January 2026, I sent the Chicago Fire leadership a fourteen-page analysis recommending an 18 million euro payment to trigger the release clause with Fiorentina. My document was built on three layers of data: pressing figures in front of the box, distance covered in extra time, and recovery frequency in transition zones. I believed Amrabat was the kind of midfielder any MLS club needed. The club's sporting director read the analysis, nodded, and rejected it flatly in a single sentence: "Amrabat has no commercial value. Nobody buys his shirt." It was an answer I had heard at least twelve times in my career, but this time it made me think for three weeks. My problem was not that the data was wrong. By the summer of 2026, Amrabat had moved to Manchester United on loan, and my old analysis circulated through professional offices, prompting a European club to contact me for remote consulting. My problem was that I had failed to understand that a correct data table can still be rejected if it is not written in the language the decision-maker needs. Correct data alone is not enough. It must be sold in the language of money and prestige the club craves. Since then, every report of mine opens with financial benefit or commercial value before moving into professional analysis. But the Amrabat case taught me a second, deeper lesson. When the sporting director rejected my proposal, he was not merely making a business decision. He was operating an evaluation system based on signals my analysis did not supply. And in the case of the empty document I received last month, I recognised the same problem: the system operates on what it believes data exists, not on what the data actually says. An empty table is more honest than a table full of wrong data Here I want to offer a potentially controversial view: the empty data table I received last month is, in one respect, more honest than hundreds of number-stuffed reports I have read in my career. Because an empty table confesses its own deficiency. A table full of false numbers, by contrast, confidently declares that it is telling the truth. In modern football and esports, this phenomenon has become an epidemic with a name: metric worship. Heat maps have become the industry's new form of divination, with vivid red-orange zones interpreted as absolute evidence of a player's operating position. But a heat map does not tell you how the player moved to reach those positions, how many passes he received at each point, the quality of those passes, or how his presence at a given point reshaped the opposition's defensive structure. The heat map conceals the player's true role in the tactical system behind a coat of eye-catching colour. I once saw a scout reject a young midfielder because his heat map showed he operated too much in his own half, lacking attacking intent. But when I reviewed seven matches of this player, I discovered he was playing in a back-three system with a specific brief: support the left flank in transitions, read defensive situations, and hold the ball until the upper line was ready. The player was not playing without ambition. He was playing his role. The heat map was not at fault; the reader was. In esports the problem is even more severe. Sample sizes in esports leagues are far smaller than in football, ranging from a few hundred to a few thousand matches per season depending on the title. Metrics such as KDA or gold share are sometimes so influenced by the patch meta that analyses built on them can reverse entirely after a single update. But because the appeal of numbers is so great, people keep drawing conclusions as if these metrics carried universal validity. Every match is a confession; my job is to read between the lines of code. But to read, you must accept that some lines have nothing to read, and that is the most important information of all. Why the sports analytics industry must learn to accept silence In my working environment, a constant pressure pushes many analysts to fill blanks with numbers rather than leave them blank. That pressure comes from three sources: media department deadlines, management expectations, and a "never say no data" culture in professional sport. An empty table sent to management can be read as a sign of laziness or unprofessionalism, whereas a table with numbers, regardless of their accuracy, is treated as a valuable output. This distorted incentive creates an ecosystem where the best analyst is not the one who draws the most accurate conclusion, but the one who draws the most confident one. And in such an ecosystem, empty tables like the document I received last month become a threat, because they force the reader to confront the possibility that they do not have enough information to act. But there is a paradox I want to point out: the people most confident in their data are also the most vulnerable when data fails. Croatia at the 2026 World Cup is a perfect example of the opposite. The American press criticised this side as "old and slow," based on average squad age and sprint speeds. But after the group stage, when I collected data from 48 matches and analysed it, I found Croatia averaged 116.2 kilometres covered per match, the second-highest in the tournament, behind only Russia. That figure appeared in no American headline. Interestingly, Croatia's average xG was only 1.08, an unremarkable number. Looking only at xG, you would conclude Croatia was a mediocre side that went far on luck. But when I analysed distance covered and the opponent deceleration model over the final 30 minutes, I found a different pattern: Croatia did not run faster than opponents, they ran further and longer. In extra time, when other teams' average speed dropped to roughly 65 percent of the first half, Croatia still maintained about 78 percent. That is why they won three consecutive knockout matches, all through extra time or penalties. I wrote a long analysis predicting Croatia would reach the final, using data from 48 group-stage matches to build a model. When Croatia did beat England in the semi-final, my piece was translated and republished by a Spanish analytics site. I received my first royalty payment of 120 US dollars, and the name Data Monk began to be mentioned by the data analytics community as a credible voice. Data is never in a hurry; it waits until you are sober enough to ask the right question. And in Croatia's case, the right question was not "How fast does this team run?" but "How far does this team run, and for how long?" The empty event of 2026: how silent stadiums taught me to reread everything The 2026 pandemic was a dark time for the entire sports industry, and for me personally. I was studying for a Master's in Sociology in Chicago, wrestling with essays on social network theory, and watching my analytics career stand on the edge of ending. When the Bundesliga returned in May 2026 with matches in empty stadiums, I decided to turn the crisis into an opportunity. I downloaded data from the first 26 matches after the Bundesliga restart and compared it with the 26 matches before in the same season. The result made me rise from my chair. Home teams won only 34.6 percent of matches after the restart, a decrease of exactly 10.4 percentage points from before. Draws surged to 31 percent. Away win rates rose significantly. And notably, average goals per match did not fall at all; football remained as lively as ever, but home advantage had vanished. I wrote a long essay titled "Empty Stadiums and the Death of Home Advantage" and posted it on Medium, never imagining how far it would spread. Three days later, the sporting director of Chicago Fire emailed me an offer to become an assistant analyst, starting with scanning GPS data for training sessions. I accepted. Two months later I was at the club's training centre on a cold Monday morning, tablet in hand, downloading data files from dozens of motion-tracking devices. That event taught me an unforgettable lesson: when the stands are empty, I see the winning formula shatter into thousands of fragments and reassemble in a different shape. Home advantage, it turns out, is not a fixed variable. It is a structural product of crowds, referees, weather, and the socio-psychological factors we habitually ignore in analysis. When one variable is removed from the equation, others rise to the surface, and a new model appears. From football to esports In recent years I have devoted a substantial share of my time to esports analysis, mainly tactical titles such as League of Legends and CS2. My experience in this field has reinforced my belief in the philosophy of interrogating data. Esports has a distinctive feature: the patch cycle is so frequent that each update creates an entirely new meta, rendering old data models obsolete within weeks. An analysis based on last season's metrics may be worthless in the current meta. In football, the basic rules have not changed for decades. The pitch is roughly the same size, the number of players is still eleven, and the objective is still to score in the opponent's goal. In esports, a publisher can change any factor at any time, reduce a skill's damage, alter cooldown timings, add or remove maps, buff a champion you never noticed. This means esports data must be interpreted with far greater caution than football data. The transfer market is only a mirror reflecting the fears of managers. And in esports, those fears are magnified by the short life cycle of game versions and the uncertainty surrounding the industry's own future. But when I look at the current period, with major tournaments approaching and clubs and esports organisations pouring money into roster construction, I see a familiar pattern. Analytical documents are produced faster than ever, data is collected more voraciously than ever, but the quality of interpretation lags behind. Decisions are still made on the basis of neat summary tables, attractive heat maps, and numbers presented without context. And behind those tables, nobody checks whether the data actually exists. Lessons from an empty document Back to the empty document I received last month. After a careful review, I sent a short report to my contact, pointing out that the empty information array was not a valid result but a symptom of failure. I proposed three actions: add an automated gate to halt the pipeline when the information point count is zero, distinguish clearly between the "no data" state and the "low risk" state in all reports, and establish a feedback process that allows analysts to say "I do not have enough information" without being seen as unprofessional. My contact replied two days later. He confirmed the problem had occurred in at least twelve other documents in the same processing batch, and the extraction stage of the pipeline had failed silently. The question that nobody in that organisation had asked before me was: "Why did we not detect this sooner?" The answer lies in the industry's own operating philosophy: too much focus on producing output, and forgetting to check whether the output has content. An empty table is not an empty output. It is a dangerous output, because it has the shape of a valid one. Progressive thought: signals for the next cycle The sports analytics industry, in both football and esports, will face a serious data integrity reckoning in the coming years. As machine learning and artificial intelligence become standard tools, the volume of analytical tables produced will grow exponentially. And as volume grows, the chance that an empty or corrupted table slips through the system grows exponentially too. The signal I am tracking next cycle is not which number will rise, but which attitude will change. Will organisations begin demanding evidence of data provenance before drawing conclusions? Will analysts be encouraged to say "I do not know" without fearing loss of credibility? Will readers learn to look at a beautiful table and ask hard questions? I do not believe in luck, but I believe in the probability of the shots that are left unrecorded. And in the modern data analytics world, the most frightening shots are not the ones that miss, but the ones never recorded at all. Those shots appear in no statistical table, are mentioned in no news report, are analysed in no document. They exist in the gaps between numbers. And those very gaps are where truth hides, waiting for an analyst sober enough to ask the right question. My next question for this season is simple: when you look at an empty data table, do you see a deficiency, or do you see an opportunity to interrogate the entire system all over again?