Between Evidence and Story: The Data Problem in Martial Arts Analysis

In the autumn of 2026, in an editorial meeting room in Tokyo, I received a three-thousand-word draft about a martial arts fight. The draft contained exactly one reliable piece of data: the names of two fighters. There were no strike counts per minute, no takedown defence rate, no round-by-round duration, not even a note about whether a fighter entered the cage with hands still wrapped. The rest was feeling — lines like "explosive fighting spirit" or "the heart of a champion." When I asked the author what his evidence was, the answer was blunt: "I watched the fight."

I tell this story not to criticise a single writer. I tell it because it exposes a larger problem: martial arts analysis rests on a far thinner data foundation than most other sports, and that fragility makes writers slide from analysis into speculation without ever noticing. Every sports story begins with a forgotten number, but when that number does not exist, what is the story woven from?

To answer that, we need to place three branches side by side — branches habitually lumped together under the single word "martial arts." They run on three different systems of logic, and any analyst who fails to tell them apart will see every later conclusion tilt.

The first branch is modern combat sport — MMA, boxing, kickboxing, Muay Thai, wrestling. Here results are decided by wins and losses, and data accumulates over time: significant strikes landed per minute, strikes absorbed per minute, takedown defence rate, knockout rate. The UFC standardised this statistical system from around 2026, letting people compare two fighters from two different decades on the same scale.

The second branch is traditional and performance martial arts — taolu, forms, routines. There is no direct opponent and no win or loss. The result is a judge's score, based on the difficulty of movements and deductions for errors. A form is graded on a points scale, not by how many times its performer dropped an opponent.

The third branch is sanda — a hybrid sitting between the first two. It has confrontation, throws, and strikes, yet it lives inside the competition system of traditional martial arts. Three branches, three rulesets, three ways of reading a fight.

A writer who folds all three into one word and applies a single analytical frame is setting a trap for himself. Based on my experience watching fights in both Vietnam and Japan, I see this mistake repeat steadily: a sanda piece analysed as though it were MMA, a taolu piece scrutinised through win-loss logic. That is an error at the level of the conceptual frame, not at the level of wording.

In MMA, when a writer says a fighter "can finish a fight," he can cite a metric. Khabib Nurmagomedov closed his career in 2026 with a 29-0 record, including eight knockout wins and eleven submission wins. That figure does not explain a whole fight, but it anchors a claim to a checkable point. Floyd Mayweather ended his professional boxing career 50-0 — a fact repeated so often it has become a label, yet still a fact.

But when analysing a taolu routine, a writer has no win or loss to cite. He has a judge's score — and a judge's score is itself a subject of dispute. A movement docked 0.1 points for "insufficient balance" is a human judgement, not an objective event. If a writer cannot distinguish these two kinds of data — hard data such as strikes landed, and soft data such as judges' scores — the analysis will blend two value systems and lose its capacity for verification.

With sanda, things grow more complex still. Sanda allows throws and grappling, but its scoring differs from MMA. A clean throw can score higher than a flurry of punches, depending on the ruleset and the event. A writer accustomed to MMA logic will default to assuming strikes are more dangerous than throws, then draw the wrong conclusion about who won on the scorecards.

This is where I recall a line I keep reminding myself of: I trust data, but I write about what data cannot measure. The problem is not that martial arts data is scarcer than football data. The problem is that people fill the gap with speculation presented as conclusion.

There is a genre of martial arts analysis I call "analysis from zero." The author receives an assignment — usually an upcoming fight or a little-known fighter — but lacks enough data to work with. Instead of saying "I don't have enough evidence," he builds a story. He personifies the fighter into a symbol, assigns him a "character" or a "destiny," then writes on as though everything were clear.

This approach is dangerous because it is not wrong at the level of wording. A piece about "fighting spirit" can read very smoothly. But it is wrong at the level of essence: it turns an information gap into a complete story, and the reader has no way to separate fact from inference.

In the extreme case, the problem lies in a near-empty input. An analysis of a martial arts piece with no title, no source, no characters, and not a single information point — nothing left but a vague label. At that point the analyst has a duty to say plainly: there is nothing to analyse. In practice, very few choose that route, because silence produces no article.

I once worked with a content production team where every pitch without data was sent back. The rule was simple: without at least one verifiable information point, nothing gets written. That rule slowed things down, but it preserved something more important than speed — the reader's trust.

In Vietnam, MMA has grown fast in recent years. LION Championship launched and runs events regularly, drawing a generation of young fighters. Trần Quang Lộc became the first Vietnamese fighter to sign with the UFC, opening a door many had thought impossible. But alongside that growth sits a data gap: many bouts in domestic events are not fully tracked, making the assessment of a young fighter depend on narrative rather than metrics.

In Japan, where I live and work, the story is different. Judo has its own scoring system — ippon, waza-ari — and karate splits into two branches: kumite, scored by points, and kata, judged as performance. A single country contains all three branches I described above. Tenshin Nasukawa, moving from kickboxing to MMA, is a living example of one person stepping across multiple rulesets — and each time, the way to read him must change.

Based on my experience watching fights in both environments, I see one notable commonality between Vietnamese and Japanese martial arts media: both are learning to use data, but at different speeds. I do not want to turn that difference into a cultural stereotype. The issue lies not with countries but with individual newsrooms, individual writers, individual articles.

But here I want to turn in another direction, because the story does not stop at missing data. Sports analysts, myself included, tend to assume that more data is always better. That is true — but only up to a point. Once a discipline has enough metrics, writers find it easy to use data as a shield against confronting what cannot be measured.

They cite numbers to avoid talking about a fighter's fear before a bout, about how many nights he slept before stepping on the scale, about the way he stands in the corner of the room when no one is watching. When everyone looks at the victory, I look for where they hide the weakness.

I spent a year interviewing fighters about competing in empty stadiums. The result surprised me: many said artificial crowd noise from loudspeakers threw off their rhythm more than it helped them focus. That is soft data — no metrics, no charts — but it explains what a statistics table cannot. When the stadium is empty, the person inside finally speaks.

This leads to a counter-intuitive conclusion. In martial arts, what is missing is not data, but honesty about lacking data. A good analyst is not one who fills every gap, but one who points precisely to where the gap is, and leaves it empty. Breaking convention does not need a loud voice; it needs evidence heavy enough — and sometimes the heaviest evidence is the sentence "I don't know."

Returning to the three branches, each demands a different handling of data. With modern combat sport, data is abundant but easily abused. A writer should use metrics to ask questions, not to close conclusions. A fighter with a high takedown defence rate does not mean he will win the next fight; it only means he usually handles being taken down. The distance between those two things is where real analysis begins.

With traditional martial arts, data is sparse and subjective. A writer must be transparent that he is reading a judge's score, not reading a truth. When a routine is scored, it is a negotiation between athlete and judging panel, and honest analysis must describe both sides.

With sanda, data is a hybrid. A writer must know where he stands between confrontation logic and performance logic. Confusion here produces not just wrong analysis, but wrong prejudice about an entire discipline.

There is one metric I often look for that few outlets publish: reaction time. Fans remember the score; I remember a fighter's expression at the thirtieth second of the final round, when he knows he is losing on the scorecards. Behind every defeat lies an overlooked decision — a stance change, a dropped hand, taking a shot to trade for position. Those decisions never appear in a statistics table, yet they decide the fight.

Martial arts is the sport where the line between evidence and story is thinner than in any other. A punch leaves no written record, a throw is not logged as a metric, a routine exists only in the scorer's memory. Precisely for that reason, those who write about martial arts must impose stricter discipline on themselves, not looser.

I believe the future of martial arts analysis lies not in how many more metrics we add, but in whether writers dare to state clearly what they are relying on. An analysis without data can still be honest — as long as it does not pretend to have data. And in an industry where everyone wants to tell a complete story, the one who dares to leave a gap may be the most serious writer of all.