The “Martial Arts” Label and the Cost of Choosing the Wrong Analytical Framework

THE “MARTIAL ARTS” LABEL AND THE COST OF CHOOSING THE WRONG ANALYTICAL FRAMEWORK

Three nights, one label

In the last three times I sat in a studio in Osaka to commentate on a martial arts event, the organisers sent exactly one line of description in advance: martial arts event. The first was an MMA night in Saitama. The next was a wushu taolu exhibition in a provincial arena. The most recent was a sanda bout.

Across those three nights, the analytical framework I had to build before going on air was not remotely the same.

For the MMA night in Saitama, I needed takedown accuracy, control time on the ground, significant strikes per minute, and finish rate before the limit. For the taolu exhibition, almost all of those metrics became meaningless: no opponent, no takedown, no rounds. There, judges score quality of movement, performance level, and degree of difficulty, adding up to a maximum of ten points. The sanda bout had its own logic: points for clean strikes, points for throws, points for forcing the opponent out of the ring, and knockouts still exist.

One label, three frames of reference. That is why I am writing this.

In my trade, labels are not an administrative matter. A label determines the very first question an analyst asks. And the first question determines almost the entire conclusion that follows.

Where the label comes from

English has a catch-all word: martial arts. Japanese, where I live and work, draws the line very sharply. means the “way of martial practice” — the path of cultivation, including kendo, judo, kyudo, karate-do. means competitive combat sports — MMA, kickboxing, boxing, professional wrestling. These two categories do not share a frame of reference, a goal, or a definition of success.

Japanese sports media does not mix them. Broadcasters allocate separate slots, sponsors allocate separate budgets, academies write separate curricula.

English merges them. And because English merges them, most of the world’s media systems merge them too. Streaming platform content management systems merge them. Advertising categories merge them. Search categories merge them. Academic injury databases for sport merge them as well.

The first consequence is obvious to everyone: someone who has just watched a UFC title fight gets recommended a taolu performance clip by the algorithm. They click, see nobody punching anybody, close it, and the algorithm records a noise signal. That noise signal then degrades the recommendation system itself.

The second consequence is far less visible but much more serious: sports medicine studies pool MMA fighters and taolu athletes into a single group of “martial arts athletes”, then take an average. That average describes no one. It does not describe the person with brain injury from cumulative head impact, and it does not describe the person with a torn anterior cruciate ligament from landing a rotational jump.

An analytical framework cannot be correct if the subject of analysis has not been classified. I did not learn this from a book. I learned it on air.

Based on my experience following bouts and performances in Japan over nearly two decades, I have settled on three subject classes. Each has its own operating logic, its own way of generating data, and its own way of consuming data.

Class one: modern competitive combat sports

This is the best-served class in data terms, and not by accident.

On the organisational side, this class includes the UFC, founded in 2026; ONE Championship, founded in 2026 in Singapore; Bellator, founded in 2026; and the PFL, founded in 2026. In professional boxing, the four major sanctioning bodies are the WBA (2026), the WBC (2026), the IBF (2026), and the WBO (2026). In kickboxing, K-1 launched in 2026 and Glory in 2026. In Muay Thai, Lumpinee Stadium opened in 2026 and Rajadamnern Stadium in 2026. In competitive grappling, the IBJJF was founded in 2026 and the ADCC in 2026.

In this class, data exists because betting money and broadcast rights money flow through it.

CompuBox began counting punches thrown and landed for boxing in 2026. The UFC published a detailed statistical database for every bout from 2026. Today, a high-level MMA fight can be recorded across more than forty variables, from significant strikes by round to control time to submission attempts and success rate.

Take Naoya Inoue as an example of how dense this class’s data is. The fighter from Kumamoto, born in 2026, has been a world champion across four weight divisions and unified the bantamweight division in Tokyo in December 2026. Every punch he threw in every round of every fight was counted and published. An analyst sitting in Hanoi can look up Inoue’s average significant strikes per round across his career without asking anyone’s permission.

That is the privilege of the first class: data exists independently of the athlete’s will.

Tenshin Nasukawa is another example of this class’s convertibility. He was undefeated across his professional kickboxing career, then moved to professional boxing from 2026. When he switched disciplines, his entire data record switched with him: the same athlete, measured by two different metric systems, and both systems exist publicly.

Zhang Weili became the first Chinese UFC fighter to win a championship belt in 2026. That event was recorded across hundreds of variables, and those very variables allow analysts to assess why she lost the belt and then regained it. No speculation is required.

Rodtang Jitmuangnon, ONE Championship’s flyweight Muay Thai world champion, is a case of intersection between Thailand’s traditional stadium system and the international media system. He has both a record at Lumpinee and an internationally standardised statistical profile.

Class two: traditional martial arts and taolu

This is the most data-empty class, and also the class I prepare for most carefully whenever I accept a commentary invitation.

The International Wushu Federation was founded in 2026 in Beijing. The first World Wushu Championships were held in 2026 in Beijing. Wushu entered the official competition programme of the Asian Games from 2026, also in Beijing. In 2026, wushu was staged as a special demonstration event within the Beijing Olympics, but outside the official medal system.

This is precisely where the analytical framework has to change completely.

In taolu there is no direct opponent. Nobody stands across from you to stop you scoring. There is no win-loss rate in the combat-sport sense. What you must overcome is a standard scoring rubric, and what you must defeat is your own margin of error.

A competitive taolu routine is scored out of ten points, split into three groups. Group A scores quality of movement, up to five points, with deductions for specific errors such as loss of balance, a missed step, a fall, or dropping the weapon in a weapons routine. Group B scores performance level, up to three points, depending on rhythm, power, continuity, and the spirit of the routine. Group C scores degree of difficulty, up to two points, based on whether the athlete performs all the declared high-difficulty elements — multi-rotation jumps, kicks that require holding balance on one leg on landing.

A taolu athlete cannot “defeat” anyone. They can only make their own total margin of error smaller than that of the athletes competing alongside them.

In career trajectory, this is the shortest-lived class of any sport I have followed. Many athletes start training between the ages of six and eight, peak between roughly sixteen and twenty-two, and retire before thirty. After retirement, almost no professional competition system takes them in. No per-bout paid league. No international image-rights contracts. No arena sells tickets to watch a form.

In injury profile, this class carries a very specific pattern of damage that the media almost never discusses: landing-force injuries. Every rotational jump ends on one leg, and the knee absorbs the entire kinetic energy of the body within a few dozen milliseconds. Repeated thousands of times a year in training, the anterior cruciate ligament, the medial collateral ligament, the patellar tendon and the Achilles tendon accumulate a debt.

And here is what I have to state plainly: an injury does not detonate in a single bout; it quietly accrues debt across many seasons. In taolu, that debt comes not from collision with an opponent, but from gravity and from the athlete’s own body.

Class three: sanda, and an unnamed hybrid zone

Sanda sits between the two classes above, and precisely because it sits between them, it is often ignored by both.

Sanda is a combat discipline with its own rule set, emerging from the Chinese martial tradition but sportified for competition. A sanda bout has rounds, weight classes, and a weigh-in. Points are awarded for clean strikes to legal targets, for throwing the opponent to the mat, and for forcing the opponent out of the competition area. A bout can end by knockout.

In other words, sanda has every element of a modern combat sport, but is governed by the wushu federation system and staged at events such as the World Wushu Championships or the World Combat Games.

The data consequence: sanda has scorecards and bout records, but almost no public performance database. No punch-counting service by the minute. No takedown-defence index. No open data source for an analyst sitting abroad to query.

This is the class a number of MMA fighters come from. When they move into the cage, their sanda record becomes a blurred gap in their competitive biography. An analyst has no way to quantify the grappling and anti-grappling foundation they bring with them.

Why the wrong framework produces a wholly wrong conclusion

I want to offer a concrete test.

Place a taolu athlete beside an MMA fighter in the same piece of analysis and ask: who has the better physical foundation? If the analyst uses “significant strikes per minute”, the taolu athlete’s value is zero. The conclusion drawn will be that the taolu athlete is weak. But that zero does not measure strength — it measures the absence of the measurement. Every number tells the truth, but a fight never tells the whole story.

Place an MMA fighter into the ten-point taolu rubric and ask: how many Group C points does he score? Unanswerable, because high-difficulty movements in MMA are not encoded as coefficients.

This is the kind of error I call a root classification error. It happens before the analyst opens any data table. It happens on the very first line of description in the organiser’s email.

I once made this error in a wushu commentary. I prepared my analysis around the concept of “pressure from the opponent” — a concept that is completely meaningless in taolu. On air, I had to correct the frame live in my earpiece, and the first four minutes of the broadcast were disjointed. Those four minutes taught me more about methodology than any book.

Extended to system scale, that same error mechanism produces three measurable losses.

The first loss belongs to research. Epidemiological studies that pool the two classes produce an average injury rate that is wrong for both. An injury prevention programme designed on a wrong average protects nobody.

The second loss belongs to the market. A merged label distorts sponsorship allocation. A sponsor seeking martial arts audiences will pour money into the class with the best data, simply because that is the class whose effectiveness is easiest to measure. This loop is self-reinforcing: the data-rich class receives more money, and the data-poor class loses ground.

The third loss belongs to the athlete. A taolu athlete who retires at twenty-five without a public data record has no negotiating material. No record, no valuation. And without valuation, no career transition.

The label and the power structure behind it

There is one detail I always check before analysing any combat discipline: who the governing body is, and what they record.

In MMA, the Unified Rules of Mixed Martial Arts were first adopted in New Jersey in the 2026–2026 period and spread from there to most US states and many other countries. The existence of a unified rule set entails the existence of state athletic commissions, and state athletic commissions entail bout records, weigh-in records, and suspension histories. In short: where there is a regulator, there is a paper trail.

In professional boxing, four sanctioning bodies coexist, plus the Muhammad Ali Boxing Reform Act enacted in 2026 in the United States, which imposes financial disclosure obligations and fighter protections. The system is complex and fragmented, but it still leaves a paper trail at every step.

In wushu, the International Wushu Federation issues a scoring code and runs competitions with judging panels. But the essential record here is a scorecard, not a bout report. A scorecard tells you the result, not the process. You cannot reconstruct a taolu routine from a scorecard. You cannot know at which second the athlete lost balance, or which jump in the routine came up short.

In sanda, there are weigh-ins, weight classes, referees, and anti-doping testing to international standards at major events. But the records are not open to the public.

This difference is not a technical problem. It is a structural one. Where betting money and broadcast money exist, there is an incentive to build a database. Where those two things are absent, records exist only at the minimum level needed to publish results.

The contrarian angle: a data gap is a market signal, not a technical failure

At this point I have to argue against myself.

The most comfortable explanation for everything above is that the industry’s classification is lazy, and that fixing the label would fix everything. That explanation fails at one fundamental point. It assumes that if classification were correct, the data would appear. But classification does not create data. Classification only creates a place to put data.

Put another way: the data gap in the taolu and sanda classes is not an industry technical failure. It is a market signal. Nobody funds taolu data collection because nobody bets on taolu. Nobody builds a takedown-defence index for sanda because no betting market demands that index.

And once seen that way, the catch-all label “martial arts” stops being a sign of laziness. It is a packaging decision aimed at maximising reach and advertising revenue. A content producer merges three classes into one category because that category has more viewers, and the ad seller needs one description line instead of three.

This leads to the professional risk I consider the most serious in the work of a sports data analyst: analytical theatre.

Analytical theatre is when you build a complete eight-dimension analytical framework — full headings, full tables, full conclusions section, full risk warnings, full glossary of terms — and all of it is built on an empty dataset. The framework looks highly professional. It complies with every formatting rule. And it contains not a single verifiable piece of information.

This is the most sophisticated form of fabrication in my trade, because it does not lie. It simply fills emptiness with form. The reader sees structure, trusts the structure, and overlooks the fact that there is no data inside.

The only defence I know is to name the gap. When data is silent, the writer must write that the data is silent. Otherwise, every average computed on an empty dataset carries the shape of truth.

But I have to push the counter-argument one step further. Even with correct classification and complete data, a conclusion can still be wrong, because data is never the fight. A metric describes a capability under conditions that have already passed. The next bout takes place under conditions that have never existed. A map is not the territory; data is not the fight.

Having said that, I need to add something about Vietnam, where I write for readers.

Vietnam has its own traditional martial arts system, with Vovinam founded in 2026 by the late master Nguyen Loc. Vovinam has appeared in the competition programme of the Southeast Asian Games, alongside pencak silat and other combat disciplines. But when I search for public performance data for these disciplines at athlete level, bout level, and year level, I find very little.

That is not because Vietnamese people do not care. It is because nobody has built a data pipeline for these disciplines. And a data pipeline is only built when someone pays to build it.

This is where I want to make three concrete proposals, at the minimum level.

One: every content product about martial arts should carry a mandatory field stating the subject class — competitive combat sport, traditional scored martial art, or hybrid discipline such as sanda. One line, placed directly under the headline or in the data description.

Two: every analytical piece should disclose the minimum data framework it uses: where the data comes from, who collected it, over what period, and how many independent sources. I still keep the habit of opening three different sources for every figure before writing, and I state those three sources.

Three: when there is no data, write that there is no data. This is the hardest line to write, because it generates no attractive headline. But it is the most honest line.

I know some will say this approach is too slow. They are right. I write slowly. I verify more thoroughly than many editors consider necessary. There are pieces I hold back for days simply because I have not yet found a third source from a market different from the first two.

But I believe this: in an industry where every party has an incentive to make the numbers look better, the slow writer is the useful writer.

What I carry with me

Forty-one years of observing this industry have taught me that martial disciplines do not operate on the same logic, and do not produce the same kind of truth.

An MMA bout leaves behind a table of variables. A taolu routine leaves behind a scorecard. A sanda bout leaves behind a short report. Three kinds of trace, three ways of reading, three levels of reliability.

An analyst has a duty to state which frame of reference they stand in before offering any conclusion. Otherwise they will apply one discipline’s scale to another discipline’s athlete, and turn their own ignorance into a number.

I still remember the feeling of those four disjointed minutes in the Osaka studio when I realised I had prepared the wrong framework. Those four minutes were not the biggest failure of my career. But they were the cheapest lesson, and I still use it every week.

For Vietnamese readers following any martial discipline, I want to leave one self-check. Next time you read an analysis of a fighter or an athlete, look for whether the piece states clearly which scale it uses. If it does not, ask yourself: which scale is being hidden, and who benefits from hiding it.

A correct label does not make an analysis correct. But a wrong label almost certainly makes an analysis wrong — and that wrongness happens before anyone opens the first data table.

Every number tells the truth, but a fight never tells the whole story. And sometimes the only honest thing a writer can offer is a gap, correctly named.