One Word, Two Rulesets: The Classification Trap in Combat Sports Analysis

On my desk in Incheon sits a folder about two knuckles thick. Inside are eight analytical frameworks any sports desk should keep ready when handling a combat event: technical-tactical matchup, fighter condition, organisational landscape, revenue model, rules and compliance, health risk, public narrative, and the industry transmission chain. Eight frameworks. Complete enough that I nearly closed the file and started writing.

Then I read the first line of the cover page. The subject label read a single word: "martial_arts". No subject class. No line confirming whether this was modern professional combat sport, traditional martial arts, sanda, or a forms routine. The eight frameworks sat there, beautiful and empty.

That morning taught me something uncomfortable about my trade: the hardest part of combat sports analysis is not missing data. It is assuming we already know which sport we are talking about.

I have never run a step in my life, yet I understand what it means to touch the finish line with your heart. But understanding with the heart will not save an analysis built on the wrong frame. And that wrong frame, in my industry, starts with one word.

Context: one word, ten rulebooks

Vietnamese calls it all "võ thuật". English calls it all "martial arts". That convenience is fine for headlines and catastrophic for analysis.

Under that wide umbrella sit rulebooks with nothing in common. Professional boxing has four major sanctioning bodies coexisting, each with its own rankings and mandatory obligations. Modern MMA runs on the Unified Rules, yet individual US state athletic commissions still license events, and each major promotion carries its own contract code. Kickboxing spans Glory, K-1, and the Thai stadium system of Lumpinee and Rajadamnern — entirely different hierarchies, where a champion in one may be unranked in another. Wrestling and sport jiu-jitsu have ADCC and IBJJF, where points reward positional control rather than finishes. Sanda has its own rules on takedowns and ring-outs. And forms competition — taolu — has no opponent at all.

Those differences are not academic. They determine every question a reporter must ask.

Take the example closest to where I live. Taekwondo in Korea exists under two schools with two rulebooks. One competes under electronic scoring, where points are confirmed by sensors in the hogu and headgear. The other is scored by human referees, with a different technical philosophy about the amplitude and crispness of a technique. An athlete from one system facing an athlete from the other, in the same tournament, reveals a set of weaknesses the other system's statistics cannot see.

I once spent an afternoon with a dataset from a youth tournament in Gyeonggi. It had win rates, bout counts, average points per bout. Enough to write something that sounded expert. But when I asked the organiser which scoring system applied — sensors or hand scoring — he looked at me for a moment and said: "It matters. Same kid can win under both. But not the same way."

Had I written without asking, I would have handed readers numbers that were technically correct and semantically wrong. That is the most dangerous error in this trade, because it leaves no trace.

The core: eight frameworks, eight questions asked wrongly when classification is wrong

When the subject class is unresolved, every analytical framework becomes a machine for producing plausible answers. Start with technical-tactical. In MMA, we discuss stylistic contrast: a wrestler against a striker gets placed in a chain of analysis — who controls distance, who pins whom to the cage, who survives two rounds to drag the fight into the fourth. In boxing, the same question revolves around punching range, footwork, and reading the opponent's hook. In wrestling and jiu-jitsu, "finish" means a positional transition, and a submission rate is not comparable to a knockout rate, even though both are called finishes. In taolu, there is no matchup chain at all — only one routine and a panel.

This leads to an uncomfortable conclusion. There is no universal metric set for all combat sports; the only shareable thing is the method of asking questions, while the metrics must change with the ruleset. Apply MMA finish rates to a boxing event, or taolu difficulty scores to a sanda bout, and you do not get mildly skewed analysis. You get a systematically skewed conclusion engine, stable enough that readers mistake it for expertise.

Move to fighter condition and career longevity. The first question is always the age curve. In sports with weight cutting, that curve is distorted by another variable: the number of cuts in a career. A younger fighter may be physiologically stronger yet weaker in accumulated damage from dehydration episodes. Cut risk must be screened first, before injury, because it is visible if you bother to look at the weigh-in log.

But for a taekwondo athlete under sensor scoring, the story changes. The centre of gravity shifts from weight cutting to force distribution and strike accuracy — recorded by devices, and bounded by the devices' activation thresholds. A visually clean technique that never crosses the sensor threshold scores nothing. That is a technical variable, not a conditioning one. Apply the wrong frame and you hunt for the cause of defeat in fitness while the problem sits in the equipment threshold.

Next, organisational landscape. Three barriers decide fighters' fates and fans' experience. The first is exclusive contracts: when a promotion holds image rights for years, the free agency fans expect does not exist. The second is title fragmentation: one division, one moment, four champions across four systems, and "who is number one" becomes an unanswerable debate. The third is cross-promotion superfights, which almost always fail for commercial rather than sporting reasons.

I once spoke with a manager in Seoul about this. He said something I wrote down verbatim: "In my division we have four belts. All four are real. But the fans only want one." That gap between four and one is the entire arena of combat sports media — and the entire goldmine of unfounded analysis.

The fourth framework is the business model. Revenue splits into four lines: gate and live event, broadcast and streaming, fighter pay, and sponsorship. The weight of those lines determines everything else. A promotion dependent on pay-per-view has a different matchmaking logic than one dependent on rights fees. And within a single event, dependence on a few big names often creates a paradox: a fighter's commercial value and competitive merit can diverge widely.

Here I want to be explicit. An analysis lacking revenue figures may not draw revenue conclusions — the silence of data is itself a form of data, and it means "unknown", not "unimportant".

Rules and compliance is where misclassification shows most. Four checks: judging, doping compliance, weigh-in and weight class, disciplinary action. The regulator may be a state athletic commission, a sanctioning body, an international federation, or nobody at all in informal challenge bouts. Once the regulator is unidentified, no penalty scenario can be built.

And this is the hardest lesson I have learned: simulating a penalty scenario with no charge on record turns speculation into analysis. I nearly did it once, in a piece about a domestic tournament. Nearly. My old editor struck the whole paragraph and wrote two words in the margin: "Not yet." Those two words saved me from an error readers would never catch — but the trade would.

The sixth framework is health and career risk: brain health, weight-cut incidents, injury, retirement security, psychological safety, systemic risk. In head-contact sports, cumulative strike counts, knockout history and concussion records are mandatory inputs for any CTE screening. In non-head-contact sports, the same question must become one about joints, backs, and ankles — injuries nobody films.

One principle belongs in every practitioner's head: the absence of a risk rating is not the absence of risk. The correct status when data is missing is "undetermined", not "neutral". Those are different things, and confusing them is how a newsroom lulls itself to sleep.

The seventh framework is public narrative. Every combat event is told through an archetype: coronation, dynasty, revenge, redemption, farewell, or crossover. Each has its own lifecycle. A dynasty story only holds with a record behind it; a revenge story only holds when there is a defeat the public remembers. With an ambiguous label, you cannot even identify the archetype — and expectation analysis without an archetype is disguised guesswork.

The eighth, my favourite, is the industry transmission chain: upstream gyms and talent supply, midstream promotions and events, downstream broadcast, data, betting and consumer markets. Transmission analysis is chain-dependent; it needs at least one actor or event as the shock origin. No origin, no chain.

And here the classification trap returns in its most serious form. If the subject is traditional martial arts and forms, the transmission chain is qualitatively different. It does not run through exclusive contracts and pay-per-view. It runs through national games policy, Olympic-bid effects, and the industrialisation of folk events. If the subject is sanda, there is yet another branch: the link between national training systems and regional professional circuits.

Three different chains. Three different maps. One word.

Contrarian angle: the problem is not missing data, it is missing classification discipline

The default reaction in my trade when an analysis cannot be written is to blame the data. "No numbers." "No sources." "No interviews." It sounds reasonable, and most of the time it is right.

Not this time. The eight frameworks were intact. The method was right. The questions were right. The only missing thing was one line confirming the subject.

That is why I think the debate about "analytics in combat sports" has the wrong centre of gravity. People argue over which metrics, which models, which algorithms. The thing that determines the validity of the whole system sits much earlier: determining the sport, the ruleset, the scoring system. Skip that, and the more sophisticated the model, the more systematic the error — because sophistication is exactly what turns a false assumption into a persuasive conclusion.

I think of fighters I have written about. People ask why I keep chasing those who finish last. My answer is partly selfish: the last finisher always forces me to ask the right question. With a champion, I can be lazy. With a fourth-place finisher, I cannot — because if I misread their rules, I will write a tribute to effort when what they needed was an explanation.

Fourth place haunts me more than any gold medal, because it is a story about resilience. But resilience can only be told when you know exactly what the person was being resilient against. A stronger opponent, an unfavourable scoring system, a sensor threshold that never triggered, a schedule of two fights a week? Four answers, four articles, four truths.

There is one more consequence worth stating plainly. When a newsroom publishes a combat analysis built on an unclassified subject label, it does not merely misinform. It installs a false standard in readers' heads. Readers learn that finish rates compare across sports, that one body's rankings say something about another's. Those beliefs outlive the article, and we will spend hundreds of pieces prying them loose.

That is a loss no table of numbers will ever show.

Takeaway: sport is a quantified dream, but it must be measured with the right ruler

The eight frameworks were not at fault. What failed was our habit of believing one word can describe a world.

Every sporting record is a reminder: limits are only whispers not yet heard. But before listening for the limit, you must know which track you are standing on. Sport is a quantified dream; I am merely the one recording it in words — and the recorder has a duty to pick the right ruler before measuring.

If those eight frameworks land on my desk tomorrow morning, I will not open the dataset first. I will write a single line at the top of the page: which sport, which ruleset, which scoring method. One line. Then all the others.