Taolu, Sanda, and the Classification Trap: How Combat Sports Analytics Is Measuring the Wrong Thing

In September 2026, at the 19th Asian Games in Hangzhou, wushu ran two events side by side on two platforms less than three hundred metres apart. On the taolu platform, an athlete performed a routine in eighty seconds, received 9.756 from the judges, and took gold. On the sanda platform, another athlete absorbed three kicks to the head, knocked his opponent down twice across three rounds, and also took gold. Vietnamese media called both of them "wushu fighters." And when I opened three different sports data platforms to look up those two names, all three filed them under a single label: martial_arts. That was the moment I understood the problem did not lie with the fighters. It lay with the label. It lay in the fact that an entire industry is applying the same measuring stick to two things whose operating logic differs at the root. A taolu athlete has no knockout rate. A sanda athlete has no presentation score. And the automated analytics system that dozens of broadcasters, news sites and data apps rely on — it does not know that. People call me a traitor, but I am only reading the current of history ahead of time. This time that current froze at exactly one point: the classification stage. Every hot take is an arrow shot into the eve of tomorrow — and this arrow is aimed at the very architecture that produces thousands of sports reports every day, including ones forty-eight pages long, with all eight analytical sections, all the tables and risk models, and not a single real fighter's name. The context here is not the story of one event. It is the story of how we define a martial art. Modern combat sports contain at least three major streams that require different methodological treatment. The first is professional combat sports: MMA, boxing, kickboxing, Muay Thai, grappling. Here the entire analytical logic revolves around wins and losses, finish rates, opponent quality, control time, durability. The second is traditional and performance martial arts, best represented by wushu taolu: there is no direct win or loss, only scores awarded on movement difficulty and performance quality. The third is sanda, the hybrid between traditional wushu and professional kickboxing that permits punches, kicks and throws. These three streams need three different analytical toolkits. Applying the first toolkit to the second produces systematic false conclusions. In Vietnam the confusion runs even thicker, because we have hundreds of traditional lineages, each with both forms and sparring content. Vovinam has forms and it has fighting. Many other schools do too. The media calls all of it "martial arts," puts it all on the same page, and sometimes scores it all against the same frame of reference. I have sat long enough in newsrooms to see the consequences. Last year a sports data company sent me a forty-eight-page deep assessment of a combat sports event. It had all eight sections, exactly matching the industrial template. It had a technical-tactical analysis table. It had a six-column risk model. It had a transmission diagram running from the upstream to the downstream of the industry. It even had a glossary of eleven professional terms, carefully explaining what taolu is, what sanda is, what the Unified Rules of MMA are, what SLpM is, what CTE is. Reading that far, I thought this was one of the most meticulous documents I had received in thirty-five years of following combat sports. Then I turned to the technical-tactical section. The "Style matchup" cell read: N/A — insufficient information. "Finishing ability": N/A — insufficient information. "Record quality": N/A. "Key metrics": N/A. I turned to the fighter-condition section. "Age curve": N/A. "Weight-cut risk": N/A. "Injury wear": N/A. "Camp quality": N/A. I turned on through organisational landscape, business model, rules and governance, health and career risk, public narrative, industry transmission. Every cell carried the same line: insufficient information to assess. And yet the report still ran forty-eight pages. It still had every section heading. It still had a block diagram drawing four organisational tiers from top to local, except all four tiers read N/A. It still had a six-row risk matrix, every row N/A. It still had a four-criterion information-value rating, each criterion scored one out of five stars with a note that there was no basis for scoring. What is frightening is not the N/A cells. What is frightening is that the template forced the analyst to fill forty-eight pages with the letters N/A rather than letting him stop at page two and write one line: "The source document contains no analysable content." That template is the architecture I am talking about. It does not allow emptiness to exist. And when a system does not allow emptiness to exist, it produces two kinds of output: either N/A spread across forty-eight pages, or invented content that sounds perfectly reasonable. Of the two, the second is far more dangerous, because it can be published, cited, circulated, and no one will notice. I saw something close to that in a television debate in March 2026, after an international MMA fight. A guest presented an analysis of "the finish rate of Vietnamese fighters" by pooling three athletes from three different disciplines: one professional MMA fighter, one amateur boxer, one vovinam fighter. He divided total wins by finish by total fights, arrived at a beautiful-looking number, and concluded that "Vietnamese fighters have a higher finish rate than the Asian average." Nobody in the room challenged the figure. Mathematically, it was correct. Martial-artistically, it was entirely meaningless. An amateur three-round boxer has no concept of a finish rate in the way a five-round or three-round MMA fighter does. A vovinam fighter works under a completely different scoring system. Pooling them and averaging is like adding the heights of three people to their weights and concluding something about their physique. This problem is not unique to Vietnam. But it is especially severe here for three reasons. First, Vietnam's martial arts ecosystem is highly fragmented: hundreds of schools, dozens of federations, many parallel competition systems with no shared database. Second, most Vietnamese combat sports data is produced by general sports desks rather than deep martial arts specialists, meaning the writers often cannot distinguish taolu from sanda, or amateur from professional fighting. Third, the pressure to produce content to an industrial template is rising, which makes filling an empty cell with N/A a less encouraged choice than filling it with a plausible-sounding sentence. The moment Eriksen collapsed, I suddenly understood that football does not need a scoreline to touch the heart. In combat sports there is an equivalent moment I remember vividly: the first time I watched an international taolu routine scored live. The judges sat in a row, scorecards in hand, eyes locked on every movement. The athlete performed a high-difficulty routine including a spinning landing on one leg. That landing was three centimetres off, and the judges deducted. No opponent. No knockout. No takedown. Just one human being and his own body, and a panel measuring precision to the hundredth of a second. If you handed that performance to an MMA analytics system, it would return a blank sheet. And if that system were not allowed to return a blank sheet, it would produce a table of very convincing numbers about a man who has never fought anyone in his life. The deeper issue is that classifying a martial art is not merely a data-engineering matter. It is a matter of money. Sponsorship contracts. Competition slots. Medical standards. Insurance policy. Injury protocol. Which protocol handles a taolu athlete's knee injury, if his entire file has been dumped into the fighting group? Under which standard is a sanda athlete with a history of concussion assessed, if his data table has no field for "head strikes absorbed" but only for "average presentation score"? These questions are not academic. They determine whether a person gets to compete, gets insured, gets to continue a career. Weight cutting is the clearest example. In professional combat sports, rapid weight reduction is the highest-mortality risk and also the most overlooked. An MMA fighter can drop ten kilograms in the forty-eight hours before weigh-in. That process brings dehydration, acute kidney injury, rhabdomyolysis, and in the worst cases collapse on the scale itself. If a data system cannot distinguish a fighting athlete from a performing athlete, it will have no field to warn about this risk. And if it has no field, it will never flag anyone. The system's silence in that case is not a safety clearance. It is merely a missing field. I have seen the same in brain health. CTE, the degenerative brain disease linked to accumulated head impacts, is studied extensively in professional combat athletes. But when you pool every martial art under one label, you cannot separate the high-exposure group from the low-exposure group. You cannot say which athletes need periodic screening, which need mandatory rest periods, which need to be removed from competition on medical grounds. A table that over-simplifies strips away the ability to protect the people who most need protecting. During a major-event cycle, when public emotion is compressed and amplified at once, these classification gaps become more dangerous. Fans follow combat sports through flags and stories. They do not care how a martial art is defined in a data field. They only care who beat whom. And content producers, under pressure to report fast, usually choose the simplest solution: one frame of reference for everything. This is where I must state clearly where I might be wrong. There is a possibility that the current coarse classification, however crude, serves a practical purpose: it gives martial arts governing bodies a single point of contact, for licensing, for event organisation. Split it too finely and the administrative system might collapse under complexity. Perhaps that generic label, despite producing analytical distortions, keeps the machinery running smoothly at the macro level. And perhaps in many cases, returning N/A across forty-eight pages — like the report I received — is actually the most honest choice the system can make. That analyst did not invent. He was merely forced by the template to write at length. I also have to concede another possibility: sometimes the confusion benefits the athletes. A famous taolu athlete, if the media calls him simply a "fighter," may reach more sponsors than if he were labelled precisely as a "performance athlete." A vovinam forms athlete might be invited onto television alongside an MMA fighter, and both share the light. Split things too cleanly and some groups may lose the attention they currently enjoy. So that generic label, in terms of short-term interest, sometimes does exactly the right job. But I might be wrong in another place too. Perhaps the problem is not at the classification stage but at the stage where classification results are used. A system can assign a coarse label and still produce sound analysis, if the reader knows to read it with context. A figure for finish rate, placed beside a note saying it applies only to professional combat athletes, will no longer mislead. Maybe the problem is not the label but the fact that the label gets stripped from its note as it travels across channels. If so, the fix is not rebuilding the classification system but rebuilding reading habits. I still lean toward the first hypothesis. Because in thirty-five years of following combat sports, I have seen too many false conclusions built on data cells that were correct but misplaced. Germany did not go home; they had been home long before the tournament began. In combat sports, the equivalent line is: the taolu athlete is not weak, he is simply being measured with someone else's ruler. And once measured with the wrong ruler, the conclusion will follow — no matter how careful the measurer. What I want to see in the next twelve months is a very specific change in Vietnam's combat sports data infrastructure. Every athlete, when entered into any system, must carry at least two mandatory classification attributes: martial arts stream (professional combat / amateur combat / performance / traditional) and applicable scoring system. Without those two attributes, the system must refuse to output any performance metric. Refuse to output, rather than output N/A across forty-eight pages. There is a big difference between the two choices, and the difference is this: one preserves honesty, the other cultivates the habit of filling gaps. The transfer market is like a detective story, and only the patient find the culprit. In combat sports, the culprit is not a person but a template. And the patient ones here are those willing to spend thirty seconds asking: which discipline does this athlete compete in, scored under which system, before entering his numbers into the table. Those thirty seconds can prevent a chain of errors stretching over years. They can prevent an athlete being misjudged, an injury going undetected, a competition slot going to the wrong person, a sponsorship contract reaching the wrong target. Above all, they can prevent an entire generation of fans from learning to read combat sports the wrong way from the start. When I think about the next generation — the journalists under twenty-five I mentor in my small programme — I always tell them this first. Before learning to analyse, learn to classify. Before learning to compare, learn to decide whether the two things are comparable at all. A hot take has value only when built on a solid definitional foundation. Otherwise it is just noise, carefully packaged. That forty-eight-page report with its eight N/A sections, I keep it in a drawer. Not for its analytical value. Because it is the most expensive lesson about the classification trap I have ever held in my hands. That trap does not lie in missing data. It lies in a system designed never to allow data to be missing. And in an industry producing content at several hundred pieces a day, admitting "I do not know" is becoming the most courageous act available. If you are reading this and about to write something about combat sports, try one thing. Before writing the first line, write three sentences: which discipline this athlete competes in, which scoring system applies, and where this data comes from. If you can answer all three, you have avoided half of the common errors. If you can answer none, do not write. Do not let a template write forty-eight pages of N/A for you, and do not let it write plausible-sounding analysis about a person who never existed. In martial arts there is a principle every fighter is taught in the first week: know where you stand before you throw. Know the distance, know the opponent's stance, know the direction of movement. Without that understanding, no matter how beautiful the technique, it is a strike into empty space. Combat sports analytics is at exactly that moment. We have plenty of beautiful techniques. Plenty of tables. Plenty of models. But we are not certain we are standing in the right place. And in a major-event season, when the pressure to deliver fast judgments is greater than ever, determining where we stand correctly will decide everything that follows.