Sports Data Analysis: The Discipline of the Null Result and the Price of Inventing a Story

Three in the morning in Osaka. The cicadas had long gone quiet, and the only sound left was the last train shrieking through the station. On my screen sat a table of eleven rows, and every row carried the same line: insufficient information to assess.

I sat with that table for forty minutes. My fingers rested on the keyboard without typing, because I knew exactly what waited behind the next keystroke. One press, and I could produce a complete story — context, characters, numbers, a climax, a conclusion — smooth enough that nobody would think to verify it. My profession draws a line at precisely that spot. On one side is data. On the other is fabrication dressed in formatting.

What people call “deep analysis” is often just surface paint over a deeper order, and this time that order told me something hard to hear: I had nothing.

An empty data table is not a failure of the tool. It is the honest testimony of the process.

The table with eleven rows

To understand why I stayed still, you need to know how an analysis file gets built. My process runs in two tiers. The first tier breaks the source text into atomic information points: athlete name, event, mark, competition date, publication source, author stance, time sensitivity. The second tier takes those grains of data and applies nine dimensions: performance and event, athlete condition, competition structure and qualification mechanism, national competitive landscape, rules and anti-doping, team and training systems, risk landscape, public narrative and expectations, and finally industry transmission.

Those nine dimensions are not ceremony. Each one is a falsifiable question. Is this mark genuine ability, or a wind-assisted run packaged as talent? Where on the career curve is the athlete, and does their peak window align with the competition window? Does the entry come from a qualifying standard or a ranking list, and which one breaks first when the schedule tightens?

The first tier that night returned an empty list. No name, no event, no mark, no date, no source. Eleven cells, eleven times “insufficient information.” When tier one is empty, tier two has no bricks to lay. And this is where most people in the trade would choose differently than I did.

The trap of an empty cell

Picture the pressure. A story is due before morning. A client is waiting. A newsroom has already promised readers something. An empty cell on screen is not harmless; it is a hole that every surrounding system is pushing you into. And the human brain is trained to fill holes.

In sports data, that filling has a polite name: “reasonable inference.” The writer tells himself he is merely extrapolating from what is known. The problem is that when there are no source data points at all, every extrapolation departs from memory, instinct, or what I call the muscle-memory template — not from evidence.

I have watched this mechanism operate at industrial scale. In 2026, when I received the SJA Young Sports Journalist of the Year award, I sat in a post-ceremony discussion and heard three colleagues recount the same match three different ways, each with statistics that “confirmed” their version. All three were coherent. All three were choosing a reading.

Data never lies; the liar is the one who chooses how to read it.

That empty table was the moment I refused to choose a reading when there was nothing yet to read.

Why tier two had to collapse completely

What is easily misread is the assumption that I was failing. In fact I was watching a process work exactly as designed. Each of the nine dimensions collapsed to the same sentence, and that simultaneous collapse is a signal, not a bug.

On performance: no event, no mark, no wind reading. A run with no knowledge of wind direction cannot be compared against a qualifying standard. A jump with no measurement point cannot separate ability from luck. No benchmark, no verdict. Collapsed.

On athlete condition: no name, no age, no personal progression curve. To say someone is trending upward you need at least three time points and a training-load context. Nothing. Collapsed.

On competition structure and qualification: no competition name, no window, no tier. You cannot discuss entry slots without knowing the meet. Collapsed.

On national landscape: no nation is named. To map strength you need names, group depth, and a talent pipeline. Empty hands. Collapsed.

On rules and anti-doping: no allegation, no procedure invoked. You cannot build a sanction scenario without a triggering event. Collapsed.

On team and training systems: no coach, no training group, no periodization signal. Collapsed.

On risk: the risk matrix has six categories, and all six sit empty because there is no subject to assign risk to. Collapsed.

On public narrative: no label to attach. No prodigy, no record chase, no comeback, no farewell. Collapsed.

On industry transmission: upstream, midstream, and downstream all carry no signal. Collapsed.

Nine collapses to the same point are not nine failures. They are one conclusion, confirmed nine times: the input source contains no analyzable content.

I learned this discipline through a mistake

In June 2026, I was invited to serve as a trial-version data commentator on a Japanese sports channel for a World Cup group-stage match in Russia. In the first half, I mispronounced a midfielder’s name three times. People could laugh, and I accepted the laughter. But the thing that kept me awake lay elsewhere: defensive-line tracking data showed the pressing structure being stretched over a short span, and I had not seen it coming, even with the tracking data in front of me.

I spent an entire month rewatching the group-stage footage. Not to fix my pronunciation, but to check what kind of signal I had missed, and why.

Mispronouncing a name is not the error; the shortfall is failing to see the outline of a system.

The lesson was not about proper nouns. It was this: when data is missing, you do not invent — you name the gap. And when data exists but goes unread, you are not permitted to pretend it is not there.

That line is the line of the profession. I have stood on both sides of it.

An empty source and two kinds of writers

There is a simple test to separate two kinds of practitioners. Hand them an empty source and demand an output. The first kind returns a full analysis page — charts, decisive conclusions. The second returns one line: insufficient information, please re-supply the source.

At a glance, the first looks more useful. Look instead at the long tail.

The first kind generates what I call cognitive debt. Every invented conclusion leaves a balance owing. It is not paid immediately, because nobody cross-checks. It accumulates. Six months later, a decision is made on that conclusion. A year later, a collective belief is built on it. Three years later, an entire strategy rests on a brick that does not exist. When the debt comes due, it does not collect from the borrower; it collects from the system.

The second kind loses in the short term. People say he lacks passion, lacks content, lacks the nerve to conclude. But he is holding a quiet asset: retrievability. Every sentence he writes can be traced back to a data point. Tomorrow, the day after, three years from now, the chain is still intact.

When everyone looks one direction, I start examining the gap behind their backs.

That night in Osaka, the gap behind everyone’s back was the eleven-row table. None of the outside sources wanted to look at it, because looking at it produces no story.

An old example that still holds

To show this process is not wordplay, go back to 2026. While new sports platforms raced to publish opinion-driven analysis, I worked for a major betting company in Osaka and published a study comparing pressing-intensity metrics across eighteen clubs in Japan’s national league.

The study surfaced something unexpected about a club in Shizuoka: their actual goalscoring rate ran well below their expected-goals figure, roughly eleven goals across a season. The media called it bad luck. I called it a structural gap through the central corridor — a side that generated high-quality chances but left openings in its defensive structure that it paid for.

The study predicted a fourteenth-place finish, not the eighth place the media praised.

The season ended exactly that way.

The point is not the number of statistics. The study was strong because every number answered one question: what does it change in the forecast? If a metric changes no decision, it is decoration.

And when a metric cannot be computed because data is missing, the honest answer is to leave the cell blank — not to stuff in a guess.

The lethal appeal of decorative numbers

This is the part that worries me most about my industry.

Readers do not verify. They react to rhythm. A piece with many numbers, many names, many upward arrows feels certain. A piece saying “insufficient information” feels hollow. So the market’s reward flows toward the fabricator, not the disciplined.

I once worked at a running magazine, where I learned the opposite over many years. Athletics teaches a writer a virtue that team sports easily erode: every statement is reducible to milliseconds, centimetres, and heartbeats. There, a wrong number is not rebutted by opinion; it is rebutted by a stopwatch.

A comeback is never a miracle; it is simply what you saw in the data three months earlier.

That is true of an athlete returning from injury. It is also true of an analysis. What you see today was already in the data. If you call it a miracle, the fault is not in the data.

Twenty-nine years and one habit

I entered the trade at a running-only newsroom, working on thousands of pieces about one narrow subject: the human foot meeting the ground. For a while I thought that repetition would wear down my ability to distinguish signal.

The opposite happened. When you watch the same motion thousands of times, you begin to hear a rhythm. You notice where the error bars sit, where the outliers are, and where a process metric is hiding a structural problem. Over the more than twenty years of athletics reporting that followed, I carried that habit into sports betting analysis when I moved to live in Japan.

Every odds movement is a heartbeat; I can only hear it with my ear pressed to the ground of data.

And when there is no ground to press an ear to, the heartbeat does not exist. That eleven-row table was a region with no ground.

The counter-intuitive angle

The familiar reaction to a null result is to call it wasted time. I think that reaction is wrong, and wrong systematically.

Consider the structure of risk in analysis. A wrong conclusion costs more than an empty one, but the costliest outcome is not knowing you were wrong. A null result is a self-reporting signal: it says the process is still intact. When a system can still say “I do not know,” it retains the capacity to learn.

A system that never says “I do not know” has frozen. It resembles a pipe blocked at the intake that still discharges product at the outlet — and people see the product, not the blockage.

This is what general audiences, including very knowledgeable sports fans, routinely overlook. We enjoy commentary because it flows. A commentator saying “I lack enough data to assert this” mid-match is read as losing rhythm. Yet that exact moment is the boundary between a presenter and an analyst.

What we call an “expert” is often surface paint over a deeper order: the ability to distinguish the known, the unknown, and the unknowable. Those three states demand three different ways of speaking. Blending them is the craft of hosting, not of analysis.

So when I refused to fill that empty cell, I was not protecting my own caution. I was protecting the accuracy of a chain that someone might pull backward years from now.

What a null result reveals about the requester

There is an under-discussed angle: a null result is also a measurement of whoever commissioned it.

When I return a page containing only “insufficient information” lines, I am implicitly asking a question. Do you want a product to publish, or an answer to use? Those differ, and they often conflict.

Someone who needs a product will push for a conclusion. Someone who needs an answer will go back to tier one, add sources, and accept a day’s delay.

In the transfer window, this pressure peaks. Transfer rumour is a current with its own rhythm: one account posts, three sports sites repost, and within two hours a possibility looks like a near-certainty. What the industry needs at that moment is a reliability filter, a verified injury status, a structural reason involving the wage bill and release clauses. All of that is data. Without data, every transfer analysis is a novel with real names.

Anchoring back to structure

There is a counter-argument I often receive: if the source is empty, write nothing, and a blank page is not analysis either.

I agree with half of it. A blank page is not analysis. But an analysis of why no analysis is possible is a product with its own value. It exposes the source’s fracture point. It states precisely what the source needs to add so that tomorrow’s analysis can happen. It sets out in advance a list of signals to watch and a trigger condition for each.

That is what I wrote in the table that night, even if I did not publish it: the article title and source, a non-empty information-point list, core viewpoints, entities involved. Four lines. Not long. But those four lines are the entire distance between real analysis and an invented story.

An era does not begin with technology; it begins with a question sharp enough to cut through the trodden path.

The question here is not elegant. It is: what does your source contain? If the answer is an empty list, then the era of analysing this text has not begun.

The price of guessing fast

One small story to close this section.

A younger colleague once asked me to review a piece about an athlete returning from injury. The piece was rich in numbers: downtime, training sessions, load increases. But when I checked, three of those markers came from three different sources, measured three different ways, and none recorded its definition. My colleague had a very solid piece. And it could not stand.

The notable thing is that he was not trying to be wrong. He simply accepted what was available because his scepticism threshold had not yet been built. A scepticism threshold is not a personality trait. It is a process, and a process takes time to form.

It took me more than twenty-five years of groundwork to build mine. And even now, whenever I pick up a new source, the first thing I do is check whether the ground exists.

Looking ahead: signals for the next cycle

The forward-looking thought I carried out of that Osaka night is not a conclusion about an athlete, a meet, or an event.

It is a question about how this industry will be measured in the coming years. As automated analysis pipelines become common, what separates practitioners will no longer be the ability to generate text. Every system generates smooth text. What separates them will be the ability to say “no,” and to state precisely why.

I am watching three signals.

First, whether sports sources begin publishing accompanying metadata — dates, units, metric definitions — or keep pushing headlines that cannot be verified. When metadata becomes standard, fabricated analysis becomes automatically more expensive to produce, and the market self-filters.

Second, whether platforms begin rewarding retrievable content. This is an economic indicator, not a moral one. As attention fragments, retrievable credibility becomes a priced asset. At that point, the disciplined stop losing.

Third, whether readers begin asking where numbers come from. This is the slowest and most important signal. A reader who knows to ask “where did this figure come from” will reshape the entire content supply chain.

And I will keep working with data tables. Some nights they are full, and I write. Some nights they are empty, and I say so. Both cases sit inside the same profession.

An empty table is not a lesson in failure. It is a reminder that an analyst’s value lies not in how many answers he holds, but in how long he stays honest with questions that have no answers yet. When the source is re-supplied tomorrow, the evidence chain starts from zero. And that is the only starting point worth trusting.