The Silent Data Table: The Boundary Between Analysis and Illusion in Table Tennis

On a December afternoon, I sat in front of my screen with fourteen data files from a domestic table tennis tournament. Over the previous three weeks, I had spent more than seventy hours collecting metrics from two hundred and thirty matches — service-point win rates, forehand loop counts per game, average rally duration. When the final summary table opened, the main data column returned a blank. Not the blank of a player who has never won a match. It was the blank of a data pipeline broken somewhere between the recording stage and the aggregation stage. In that moment, an old question returned: when there are no numbers, what should an analyst do? I could sit down and construct a plausible-sounding story about player form, assigning strengths and weaknesses no one could verify. Or I could admit that I was standing in front of a void, and write about that void itself. I chose the second option. Not because it is easier — it is far harder — but because it is the only honest way to practise this craft. Vietnam's table tennis data analysis sector sits at a strange intersection. While football has long had standardized advanced metrics, table tennis still fumbles with the crudest numbers. The International Table Tennis Federation publishes its rankings weekly, but that is just accumulated points — a summary table, not the full testimony of a match. At the domestic level, the National Sports Festival and the national championship still mostly record win-loss outcomes by game, lacking any micro-level data layer for individual rallies. This creates a paradox. We have many matches, many players, but very little data that can be used for serious analysis. And when data is scarce, the pressure to "fill the void" becomes enormous. That is precisely the moment when an analyst is most likely to lose themselves. I recall the lesson of 2026, when I used an expected-goals model to show that Germany — the reigning World Cup champion — risked elimination from the group stage. After their loss to Mexico, I calculated Germany's expected goals against over their first two matches at 3.2, while their attack generated only 1.8 expected goals. I published a bold claim and was ridiculed severely. When Germany lost 0-2 to South Korea and were eliminated, I received thousands of apologies on social media. But the real lesson was not that I had been right. It was that if the data had been empty that year, I could not have written a single word. And the worst thing I could have done then was to invent a plausible-sounding story. Back to the empty table in front of me. In table tennis, every match is a sequence of decisions unfolding in a far shorter time than in football. An elite rally lasts an average of seven seconds. A game can end in four minutes. That means the analyst's observation window is very narrow, and the margin of error in observing with the naked eye is very large. Without data recording each point precisely, people easily brand a player "mentally weak" just because they lost the last three points — when in reality their clutch-point win rate across the season may be the best on the team. That is why I built the habit of recording raw data for every match I follow. For each encounter, I log four metric groups: service-point distribution by zone, win rates on the third and fifth balls, conversion rate from defence to counter-attack, and win rate on deciding points. These four groups are not a complete system. But they are enough to produce a picture that can be verified, rather than one painted by feeling. I learned this method from a failure. Years ago, I wrote an analysis of a young player hailed by the media as a "prodigy" after he won three consecutive matches at a youth event. I had no detailed data, only win-loss results. I wrote that the boy had "superior competitive courage" and "all-round technique". Six months later, when he met stronger peers, he lost four of five matches and dropped out of the top group. I realized I had not analysed — I had told a story. And that story had no basis beyond three wins. Since then, I have set a principle: never draw conclusions about long-term form from a sample smaller than thirty matches. Three consecutive wins may be a signal of talent, but they may also just be a verifiable probability streak that anyone encounters in their career. The difference between a great player and an average one is not whether they have a winning streak, but the frequency with which they repeat it across tournaments, under different conditions. In Vietnamese table tennis, this problem is even more acute. Domestic tournaments often have limited numbers of matches, and key players sometimes play only a few top-level matches a year. This makes analysis based on large samples an almost insurmountable challenge. The analyst is forced to work with small samples, and with small samples they must be extremely careful with every conclusion. That is why I always tell younger colleagues: numbers do not lie, but people who read numbers do. A 70 percent win rate sounds impressive, until you know it was calculated over ten matches against opponents all ranked outside the world's top fifty. A 60 percent service-point win rate sounds good, until you realize the opponent cannot return a sidespin serve properly. Context gives a number its meaning, and ignoring context is the fastest way to turn analysis into self-deception. When my data table is empty, I have no context. I have nothing. And in that situation, the most honest action is to say plainly: I cannot analyse this. I can describe a match by eye — who loops harder, who moves faster — but that is observation, not analysis. Analysis needs data, and data needs disciplined recording. There is an interesting thing I learned working with table tennis data: the shortest moment in a match often contains the most truth. A rally lasting two seconds after the serve — usually called the third ball — reveals a great deal about a player's level. If they win many points at this stage, it shows their ability to read spin and make attacking decisions is very good. If they frequently lose points here, it shows they are being pushed into passivity by high-quality serves. But if no one records the third-ball win rate, that information disappears. And when information disappears, people revert to emotional stories: "this player has nerves of steel", "that player lacks consistency", "that player is mentally weak". These judgements sound very certain, but they have no basis for verification. They are myths built on a foundation without numbers. I have been criticized for being too dependent on data. A colleague told me that table tennis is a sport of feeling, of moments of transcendence, of rallies that cannot be measured by numbers. I agree in part. No metric measures the moment a player, facing elimination, still dares to launch a decisive loop. But I disagree with the rest: precisely because there are moments that cannot be measured, we need data all the more to determine how much those moments are truly worth. When the stands are empty, I see the truest player. No cheering, no crowd pressure, only pure technique and data undistorted by the masses. In an internal tournament without spectators that I once followed, a player once considered "only strong when cheered" won three consecutive matches with a third-ball win rate of 68 percent. That number was the opposite of the story the public believed about him. But it matched exactly what the data had recorded about him over the previous two years. The truth was: he had always had a good technical foundation, but crowd pressure had distorted his image in the eyes of viewers. This is why I believe raw data is the true testimony of a match. The ranking table is only a summary, and a summary is always edited. A player ranked twentieth in the world may have better micro-metrics than one ranked fifteenth, but be less fortunate in decisive matches or play less due to injury. If one looks only at ranking, one will miss that truth. If one looks at raw data, one sees it immediately. In the context of Vietnamese table tennis, where micro-data is still scarce, building a standard recording system is priority number one. This does not require expensive technology. It requires discipline. A coach can record a student's service-point win rate in every training session. A journalist can record the point distribution of every player in every match they follow. A fan can record the clutch-point win rate of favourite players. After a few months, these small notes will form a verifiable database. But I must admit one thing: even with complete data, I still cannot tell the whole story of a match. Data tells me what happened, but it does not always tell me why. A player may lose a point due to a wrong tactical choice, or an ankle injury, or a sleepless night. Data records only the final result, not the full set of causes behind it. That is why I say correlation is not causation. A player with a high service-point win rate usually wins many matches, but that does not mean good serving is the sole cause of victory. They might win because they serve well, or they might serve well because they are playing confidently, and that confidence comes from other factors like fitness, morale, or a weaker opponent. Data cannot perfectly separate those factors. And anyone who claims they can is selling you an illusion. This is where I differ from many in the industry. I believe in data, but I do not believe in data as a religion. Data is a tool for asking better questions, not for giving absolute answers. When I say "this player has a 62 percent clutch-point win rate", I do not say "this player will win the next deciding game". I only say that, based on history, that probability is real. And probability never guarantees the outcome of a specific match. Back to the empty data table. I decided not to write an analysis of that tournament. Instead, I wrote a three-page internal note about what had happened to the data pipeline. I stated clearly: the on-site recording stage failed in seventeen of two hundred and thirty matches; the data synchronization stage between volunteers had format issues; and the final aggregation stage lost data due to an error in the automated processing workflow. This was not a failure of analysis. It was a failure of engineering. And that difference matters. If I had merged the two together, I would have created the illusion that table tennis analysis is impossible. But the truth is: table tennis analysis is very possible, provided the data is collected correctly. The problem is not the sport; it is the infrastructure. This leads me to a broader observation about the sports data industry. We live in an era where data analysts increasingly penetrate the dressing room, the tactical meeting room, the coach's decision-making process. This has good and bad sides. The good side is that decisions become better grounded. The bad side is that analysts sometimes draw conclusions disconnected from the match's real rhythm — for example, advising a player to change serve tactics based on a model, without accounting for the player being tired or suffering from shoulder pain. I witnessed such a situation at a domestic tournament. A coach was given a report showing his student should attack more into the opponent's left zone. The report was based on data from seven previous matches, in which the opponent had a left-zone defensive rate twelve percentage points weaker than the right. The coach trusted the report and instructed his student to hit into the left. But the opponent in that match adjusted, stood further right, and returned every left-zone shot. The coach's student lost two straight games before returning to the old tactic. The data had been correct when it was collected, but it could not predict the opponent's adaptation. This is the limit of every model. They are based on the past, and they assume the past will repeat. But in elite sport, every opponent is changing, learning, adapting. A good model must have an adjustment variable for that change. And the best adjustment variable is not a number — it is the eye of a coach who understands his student and his opponent. I say this not to deny data. I say it to stress that data and intuition are not opposites. They complement each other. Data tells you the trend; intuition tells you when that trend will change. A good analyst knows when to trust the number and when to doubt it. There is a sentence I often remind myself of when starting a new analysis: if I did not want to provoke controversy, would I still write this piece? If the answer is no, I should re-examine my motives. Going against the current is only valuable when it is based on evidence. If it is based only on a desire for attention, it will soon become a performance, and the performer will be exposed. In the case of the empty data table, I could not go against the current in a compelling way. I could not write that "player X is in terrible form" or "tactic Y is outdated" just to spark debate. I had no basis for that. And without basis, every claim is just air. This is perhaps the hardest lesson I learned in eighteen years in this craft: knowing when to be silent. Young analysts often want to prove they know a lot. They write a lot, conclude a lot, predict a lot. But over time, I realized that an analyst's value lies not in the number of conclusions they offer, but in the quality of evidence behind those conclusions. And sometimes, the best quality of evidence is admitting the evidence is not enough. I think about this every time I begin a new data cycle. In eighteen years, I have witnessed many such cycles: from a time when table tennis data was recorded only in notebooks, to when analysis software began to appear, to when advanced metrics became standardized. Every cycle brought new insights, but also new illusions. And the greatest illusion of the data age is believing we can measure everything. We cannot measure courage. We cannot measure the moment a player decides not to give up. We cannot measure the feeling of a training session when everything clicks. But we can measure things close to them: clutch-point win rate, conversion from defence to counter-attack, the frequency of holding a lead. These metrics are not perfect, but they are shafts of light in a dark room. And for me, humility before the limits of numbers is not a sign of weakness. It is a sign of maturity. Young analysts often believe data can answer every question. Older analysts understand that data can answer only some questions, and that the most important question — who this team or player really is — often lies in a region data cannot reach. Back to my abandoned article. I did not publish it. I sent an internal note to the editorial board explaining the infrastructure failure and proposing remedies. Some colleagues were surprised I did not write an analysis. Some thought I had missed a chance to create a stir. But I knew I had done right. Because if I had written an analysis based on empty data, I would not have deceived only readers. I would have deceived myself — and that is the worst thing an analyst can do. I do not know whether that tournament's data pipeline will be properly repaired in the future. That depends on factors beyond my control. But I will follow it. I will keep recording. And at some point, when the data is thick enough to tell a verifiable story, I will write the first real analysis of that tournament. That is the only way I know to practise this craft decently. The question I leave for myself and for those in Vietnamese sports data analysis is this: when data is empty, do we dare to be silent? Or will we always be tempted to fill the void with plausible-sounding stories? The answer to this question, I think, will decide whether Vietnamese table tennis analysis can build genuine credibility. Because an analyst's credibility comes not from what they claim, but from what they refuse to claim when they have no basis.