Combat Sports Analysis With Empty Data: The Biggest Risk Is Inventing Conclusions

On the morning of August 13, 2026, in Binh Duong, I opened a twelve-page analysis file about a martial arts article. The Information Points field — the list of raw facts extracted from the source text — was empty. Not one name. Not one figure. Not one event. Article Title read N/A. Article Source read N/A. The only label that survived the entire pipeline was martial_arts.

Twelve pages. Zero data points.

What made me stop was not the emptiness itself but how those twelve pages handled it. They did not collapse. The eight-dimension analytical framework stayed fully intact: tables, a transmission diagram, a risk matrix, everything in place. The only difference was that every cell carried the same line: "N/A — insufficient information."

Twenty years in this trade taught me one thing. A table stuffed with N/A is more dangerous than an empty table. An empty table forces you to go out and find data. A table full of N/A gives you the feeling that the job is done.

In sports media, that feeling is the most expensive thing there is.

Speed has overtaken verification

Over the past three years, sports newsrooms across Southeast Asia — including small operations in Ho Chi Minh City, Hanoi, and Jakarta — have rolled out automated content pipelines in unison. The common architecture has two stages. Stage one decomposes the source article into information points: fighter names, events, organizations, dates, figures. Stage two uses those points as raw material to generate deep analysis against a preset framework.

The architecture is technically sound. It breaks somewhere else: stage one is usually the only stage nobody checks.

I once witnessed a variant of this problem at a much smaller scale. In 2026, while working as a mid-level staffer for a digital sports platform in Binh Duong, I was assigned to film the national youth athletics championships. In the women's 200 metres, a sixteen-year-old athlete from Dong Thap named Nguyen Thi Thanh Van finished twelfth. The results sheet said she lost. But the frame data showed that in the first 0.01 seconds after the start, her stride frequency was the fastest in the entire field.

I proposed changing her drive-phase technique. A veteran coach laughed out loud: a woman pretending to understand biomechanics. I did not back down. I collected more data from three slow-motion sessions. Six months later, Thanh Van broke the national youth record with 23.48 seconds.

The lesson sits right there. A results sheet is one layer of data. Frame-by-frame footage is another. Someone who only reads the results sheet will never see starting speed. GPS numbers do not lie; only the people reading them do.

When both layers of data are missing, the most honest reader has to say: I do not know.

Based on my experience tracking matches and championships over many years, I have noticed a recurring pattern. Newsrooms invest heavily in stage two — the presentation layer, the storytelling layer, the formatting layer — because that is the part readers see. Stage one is quiet, dry, and ungratifying. And precisely because it brings no praise, it gets abandoned.

Empty and neutral are two different things

The August 13 file concluded that no analytical judgment could be issued. That conclusion was correct. But it came with the most important sentence in all twelve pages: the correct status must be called "unknown," not "neutral."

The distinction is not wordplay. In combat sports, "neutral" is a judgment. If I write that a fighter has neither an advantage nor a disadvantage, I am asserting something about that person's ability. I am saying: conditioning is fine, technique is balanced, mentality is normal, no sign of decline.

But if I have never had data on strike absorption, weight history, or camp days, I am not allowed to say "balanced." I am only allowed to say "unclear."

In medicine this principle has a name: absence of evidence of risk is not evidence of absence of risk. A patient who has not been scanned is not a healthy patient. He is an unscanned patient.

The entire risk matrix in that analysis file — brain health, weight-cut risk, accumulated injury, post-retirement security, psychological safety — was left blank. The reason was not that no risk existed. The reason was that no one was named, so no risk could be attached to anyone.

That is the point I want to press hardest. Silence in the data is not proof of safety.

The classification problem: which framework for which subject

One technical detail in the analysis file deserves its own dissection, because it explains why filling data in carelessly is more dangerous than leaving it blank.

The domain label of the source article was martial_arts. But the analysis pointed out that this label is not enough to select an analytical framework. Martial arts contains at least three classes of subject that differ in nature, and each requires its own logic.

The first class is modern competitive combat sports: MMA under the Unified Rules, professional boxing, kickboxing under K-1 or Glory, Muay Thai, grappling under IBJJF or ADCC, sanda. For this class, the correct framework is wins and losses, finish rate, striking accuracy, takedown defence, submission frequency, and defensive grappling efficiency.

The second class is traditional and performance martial arts. For this class, the logic is entirely different. There is no direct opponent, no rounds. The measures are difficulty score, performance quality score, movement stability, landing error, and posture accuracy.

The third class is the federation and governance layer: state or national athletic commissions, and international federations such as IMMAF, WAKO, and IWUF.

Read that point carefully. If I apply a wins-and-losses, finish-rate framework to an article about performance forms, I will be wrong in a systematic way. Every conclusion will drift in the same direction. Readers will believe it, because the presentation structure looks professional. The tables look complete. The numbers look specific.

This is the most frightening failure mechanism in automated analysis. It does not invent numbers from nothing. It applies exactly the wrong framework to exactly the right subject.

I have met a real-world variant of this error. In 2026, at the World Cup in Russia, during the opening match at Luzhniki Stadium, the host nation beat Saudi Arabia 5-0. I called midfielder Samedov "Semidov" three times in the first half. Social media tore into me. One person even suggested I should go write about fashion instead.

I did not delete the post. I published a public apology, then spent thirty days rewatching forty broadcast matches to record how each player's name is pronounced locally. I built my own glossary with cultural notes and tested it from the quarter-finals onward.

When I mispronounced a player's name, I learned how to listen to the match. Getting one name wrong is less dangerous than applying a wrong framework to an entire career.

Eight analytical dimensions and the cost of filling in blanks

The August 13 file preserved all eight dimensions: competition and tactical analysis, fighter condition and athletic longevity, event and organizational landscape, business model and market, rules and governance compliance, health and career risk, public narrative and market expectation, and finally industry transmission.

Each dimension has its own framework. The tactical dimension has a four-metric table: style matchup, finishing ability, record quality, key metrics. The condition dimension has four rows: age curve, weight-cut risk, injury wear, camp quality. The business dimension has four revenue rows: PPV and broadcast, gate and live events, fighter pay, sponsorship. The compliance dimension has four check items: judging and scoring, drug-testing compliance, weigh-in and weight class, disciplinary action.

All of it was blank. And all of it was correct in being blank.

Now imagine an undisciplined pipeline. It has four empty cells in the condition dimension. It needs to fill them so the output looks complete. What will it do?

It will take the average sample of fighters in the same weight class at the same age. It will assign "average risk" to the age curve. It will assign "no abnormal signs" to weight-cut risk. Those sentences sound harmless. But they create a person who does not exist.

And when readers encounter a profile with all eight dimensions filled in, they tend to believe it. Cognitive research shows that people judge the credibility of information largely by its presentation. Balanced tables, clear headings, technical terminology — all of these are trust signals, regardless of whether the content inside is real.

That is why I say: the most serious error in automated analysis is not blatant fabrication. Blatant fabrication is easy to catch. The serious error is filling a blank with something that sounds reasonable.

Every play is a hypothesis, and I am the one who likes to test it. But testing requires a ball to test. No ball, no match.

The minimum list: five things that cannot be missing

If I had to extract one tool newsrooms can use immediately, I would propose a minimum list of five items. Missing any one of them, the pipeline should stop itself instead of speculating.

First, the source article's title, source, and publication date. Without these three, neither freshness nor source quality can be assessed. In the August 13 file, both the title and the source read N/A. That means there is no way to trace any claim afterwards.

Second, a determination of the subject class. This is the most mandatory and most frequently skipped item. Modern competitive combat sports, traditional performance martial arts, or sanda. Three classes, three frameworks, three different sets of metrics. The martial_arts label is not enough.

Third, at least five raw information points with concrete facts — that is, proper names, events, organizations, dates, figures. Figures must carry units and sources. Below this threshold, any analysis is decorated fiction.

Fourth, a list of extracted entities: fighters, coaches, gyms, events, organizations, governing bodies. This is the backbone of any transmission diagram. Without a source entity, no transmission diagram can be drawn.

Fifth, and this is the item I value most: mark clearly what is a fact taken directly from the source article, what is an inference made by the analytical layer, and what is a low-confidence speculation. Three tiers, never blended.

My reason for ranking item five first in importance: readers have a right to know what they are reading. A speculation labelled as speculation still has value. A speculation presented as fact is a lie, even when the writer did not intend it.

The contrarian angle: an empty file may be the most honest document in the newsroom

This is where I want to go against the crowd.

The default reaction on seeing an empty analysis file is to conclude the system is broken. At the technical layer, that conclusion is correct: stage one failed with three specific defects — no information-point extraction, no entity extraction, no subject-class determination.

But seen from the editor's side, the August 13 file did one thing that a great deal of Vietnamese sports writing gets wrong every single day: it refused to conclude when there was no basis for a conclusion.

I read enough to know the density of blank-filling content in this industry. A fighter wins three in a row and immediately there appears an "analysis" about a steel mentality, about an unbreakable spirit, about extraordinary growth. Who those three wins came against, by what method, what form the opponents were in, what the ruleset was — those questions usually go unasked.

The cause is structural. Most combat-sports content in small markets is generated from promoters' press releases. The release comes with characters, storylines, and angles ready-made. The writer only has to restructure it. Nobody has to leave the office. Nobody has to rewatch footage. Nobody has to call and verify a single figure.

That is also why I have told young people in this trade repeatedly that the transfer race between big clubs is largely a brand arms race. The genuinely valuable signing sits at a small club, where a foreign player slot is measured by data rather than by the fame of the name on the shirt. And you only see that if you are willing to open the data and read it.

Journalism that only reads press releases does not produce informed audiences. It produces audiences who know the story by heart but do not understand the match.

So the empty August 13 file, in a very narrow sense, is an honest document. It says: I do not know. In an industry where saying "I do not know" is treated as failure, that is a rare act.

But honesty does not mean usefulness. An empty file correctly states that there is nothing to analyse yet, but it does not give the newsroom something to analyse. The next job is not to publish — it is to go back to stage one and fix the three defects there.

Honesty is a necessary condition. It is not a sufficient one.

Closing: what I want newsrooms to write on the wall

I do not think the biggest problem in Vietnamese sports journalism is a lack of technology. I think the problem is that we are using technology to accelerate a process that was never trustworthy to begin with, and because speed went up, we assume quality went up with it.

A pipeline can process two hundred articles a day. That does not mean it understands two hundred matches. Output volume is not a measure of capability. It is only a measure of bandwidth.

On August 13, 2026, in Binh Duong, I closed the analysis file without publishing a single line from it. I sent a message to the pipeline owner: stop the run, fix stage one. Below it I added a line I think should be printed and taped to the wall of every sports newsroom: when the data goes silent, the most decent writer is the one who goes silent with it and goes looking for sources, not the one who speaks on its behalf.

If you are running a similar process, check your stage one this week. Do not check the output. Check the input. The output will always look fine — that is the nature of formatting. Only the input dares to tell the truth.