When the Analytics Board Returns a Blank Page: The Limits of Automated Data in Martial Arts

An analytics board appears on the screen. Every cell is empty. Only one label stays lit: “martial arts”. No fighter’s name, no strikes, no injury data, no tactical notes. An automated analytics system ran its entire process and returned a single word.

For anyone whose job is reading sports data, that scene is familiar to the point of being frightening. It is like opening a fighter’s file hundreds of pages thick, flipping through all of it, and discovering every page is identical: blank. The system ran, but it did not understand. It completed the shell of an analytical process and forgot the whole interior.

In martial arts, where every strike is a survival decision made in a few hundredths of a second, automated analysis is becoming an industry. Motion-tracking platforms, force sensors, high-speed cameras and machine-learning models are brought into the gym. The promise is enormous: predict strikes, measure impact load, warn of injury. But when a system returns a blank page, the real question is not “where did the machine break”, but “what did we expect wrongly”.

When martial arts data goes hungry

Martial arts is among the hardest sports to digitize. Football has a ball, a pitch, sidelines, a steady running rhythm for algorithms to latch onto. Martial arts does not. Two fighters occupy a tight space, movement is non-linear, and impact happens at a speed cameras often cannot follow. A taekwondo kick can last under two tenths of a second. A judo throw changes the axis of rotation of two bodies in an instant. A boxing hook may leave a trace on a sensor for only a few frames.

Because of this, martial arts analytics systems are usually built to work with dirty data: multi-angle video, uneven lighting, low frame rates. But being built to tolerate dirt does not mean tolerating emptiness. When the input is empty, every clever design is meaningless.

There are three forms of data starvation I see often. First, the input is in the wrong format: a video clip, a scan, a page locked behind a paywall, and the system receives an empty string. Second, the extraction process fails but is swallowed silently, leaving default values like “unclassified” or “undefined”. Third, the text is too long, too short, or not in a Latin script, so the extractor returns a zero.

In all three cases, the result is an analytics board that is beautiful in form, full of cells, but contains not one piece of information. And the danger is that many people will read that board as if it were a conclusion. A blank board does not shout that it is blank. It just stays silent, and silence is easily mistaken for agreement.

It is no accident that martial arts such as judo, taekwondo and karate appeared later in the data-analytics wave than football or basketball. While football has had decades of player-tracking data, martial arts still relies largely on the coach’s eye. Each time a new system is introduced, expectations are pushed higher, and each time the system returns a thin result, trust drops a notch.

What a genuinely deep analysis requires

To call a martial arts analysis deep, it must answer four questions. One, does the system extract the core claims, meaning does it know what this fighter did, in which round, at what frequency. Two, does it recognize the writer’s stance, whether the piece is praising, doubting, or selling an idea. Three, does it map the entities, who fought whom, which coach, which event, which injury relates to which injury. Four, does it capture the source and the timing, when the number was published, by whom, in what context.

Missing any one of these four, the analysis slides out of the deep zone and falls into the decorative zone. A scorecard without claims is not analysis, it is an empty frame. A chart without a stance is not an opinion, it is an illustration. A list of entities without sources is not data, it is a rumor presented neatly.

The body does not know how to lie, but data needs someone who knows how to listen. This is what I always remind myself when sitting before any data board. Numbers do not speak on their own. They speak only when someone places them in the right causal chain.

In fifteen years of watching sports, I learned that a good analytics system is not the one returning the most numbers, but the one that knows to stay silent when it has nothing to say. When a model returns a blank page, that can be an honest signal: it is admitting it does not understand. The problem only arises when we force it to speak, and it begins to fabricate.

I still remember the feeling the first time a model I built returned an empty result. I checked the source code three times, thinking I had mistyped. But the error was not in the code. It was in the fact that I had assumed the data existed, when in reality I had never confirmed it.

I once witnessed a technical meeting before a major fight. The coaching staff laid on the table a thick report printed from an automated analytics system. The report had every section: injury history, striking trends, moments of lost focus. But read closely, every section pointed back to the same single source: a three-minute video online, with no context, no date, no event name. The whole report was built on such a fragile fragment.

That was when I understood why I am allergic to beautiful analytics boards. What we call data is often just an uninvestigated fragment. An uninvestigated fragment, presented in a neat table, will dress itself in the appearance of a fact.

There is one detail I always check before trusting any martial arts analytics board: the date. A number without a date is a floating number. It may be true of a fight three years ago, and completely wrong for last week’s fight. In a sport where one injury can change the entire striking mechanism, a number not tied to a moment is a meaningless number.

The trap of automated analysis

This is where I want to speak plainly. The sports-analytics industry is entering a phase where data analysts move closer to the locker room than ever, yet the gap between the number and the body’s actual rhythm has never been wider.

An algorithm can count strikes precisely, but it does not know that the thirty-seventh strike in a fight is the one where the knee begins to drift off axis. It can measure speed, but not know how much of the shoulder joint’s remaining range of motion that speed was paid for with. It can draw a beautiful fatigue curve, but not know how many hours the fighter slept before the bout.

The problem is not that the data is wrong. The problem is that the data is read without the body that produced it. A dense schedule does not just tire players; it signs its name on every body. A model with no concept of that signature will always mispredict in the same way.

There is a paradox I see in almost every martial arts analytics system: they are judged by how many data cells they fill, not by the quality of what they say. A board with a hundred filled cells is considered better than one with ten filled cells, regardless of whether those ten are worth a hundred times more. That very way of judging has quietly encouraged fabricating data instead of admitting emptiness.

I do not oppose automated analysis. I oppose making it the final verdict. A system returning a blank page is less dangerous than one returning a page full of numbers born from nothing, because a blank page forces us back to the human. A full page lulls us into the feeling that everything has been understood.

Martial arts, with its non-linearity and speed, is the harshest test for any analytics system. If a model cannot handle a kick under two tenths of a second, it will not handle an injury lasting a whole season. And if it cannot extract a single claim from an article, do not expect it to extract a diagnosis from a body.

Takeaway

The problem of the blank analytics board is not a technical bug to fix. It is a reminder that before asking where the machine went wrong, we should ask what we placed in its hands. A system can only be as deep as the data it is fed.

For those in sport, the lesson lies elsewhere. Do not let a neat table replace sitting down and watching the body. Before being a player, he is a survival question. And no algorithm can answer that question for us.

What I want to see next season is not a smarter analytics system, but readers of data who know when to fold the table away and look straight at the fighter. Because in the end, what we call bad luck is often just an uninvestigated fragment, and what we call data is sometimes only a blank page, carefully framed.