Nine Empty Sections and One Temptation: Data Discipline in Sports Analysis

On the third night in Surabaya, I reopened a report whose framework I had built myself. Nine sections. A transmission map. A risk matrix. A patch-analysis section. A tournament-system section. A club-finance section. Every cell contained text, and every cell said the same thing: insufficient information to conclude.

The source document attached to it contained nothing. No game title, no patch number, no team, no player, not a single metric. Only the skeleton, standing there, complete enough to look trustworthy.

What kept me up until two in the morning was not the gap. It was the very familiar feeling of a hand resting on a keyboard: one plausible inference, and this skeleton would be full within twenty minutes. I knew exactly what it would be filled with, because I had done it before.

In this profession, the workflow usually runs in two stages. The first stage extracts from the source text: who the article is about, which tournament, which dates, who is being quoted. The second stage is where a specialist reads the extracted data and issues a judgement. It sounds dry, but it is identical to how a scout works: watch the tape first, take notes first, conclude later.

Football and esports share the same logic. A decent opponent-review session begins by establishing the competition version, the format, the schedule, who starts, who is returning from injury. Skip the fact-establishing stage, and everything downstream is emotional interpretation dressed in terminology.

The problem is that the extraction stage can return an empty result. Empty in the literal sense: a list of information points with zero elements. A writer who receives such a document usually faces an unspoken choice. Either stop and report that there is nothing to analyse. Or fill it in themselves.

I once chose the second. In 2026, while working as a data coordinator for Surabaya United in Liga 1, I reported that my team had 63 percent possession in the first half against Persib Bandung and recommended pushing the defensive line higher. We lost 0-3, conceding twice into the space behind the full-backs. Three nights later I reviewed every passage of play and found what I had missed: the opponent's PPDA showed they had deliberately surrendered the ball to draw us up the pitch and counter. I sat down and wrote a ten-page self-critique, sent it to the coaching staff, and proposed a mandatory cross-check procedure before every match. The mistake in Surabaya taught me to question data, not to trust it.

Three failure modes appear when an analyst meets an empty input, and all three already exist in Vietnamese and regional football, just wearing different shirts.

Silent subject substitution. This is the most dangerous kind, because it leaves no trace. The writer does not say "I do not know which game this is"; the writer picks a plausible game, assigns a version, assembles a roster, and writes fluently. To a reader, the analysis reads convincingly. To a cross-checking specialist, it is wrong from the root.

In football, this type has a more familiar name: the unsourced transfer rumour. An account posts that "Club A has agreed personal terms with Striker B", accompanied by an old photograph. Larger outlets pick it up and add "reportedly". Three days later the story vanishes without a correction. Nobody verifies it, and nobody is accountable for the subject they invented.

In esports, the type is subtler. A team is assigned a "map-control playstyle" while the live patch has completely changed the tempo of matches. The claim sounds expert, but it is anchored to a version that no longer exists. What is notable is that the writer is usually not deliberately lying. They are reacting out of a reflex the industry has cultivated: every question must have an answer.

The illusion of framework completeness. A report with full headings, full tables, full charts creates the impression of careful work. But formal completeness does not equal the existence of content.

I have seen widely shared advanced-metric leaderboards where the poster names no sample, no collection window, no tournament separation. With a small sample, one anomalous match is enough to skew the entire ranking. When collection conditions change, the metric loses any ability to be compared over time. A table packed to the brim can still be hollow, while a cell that plainly reads "insufficient information" cannot pretend to have been filled. The difference between the two is the entire value of the profession.

Germany at Euro 2026 is the example I still use in arguments. They generated a volume of chances far beyond their opponents in the knockout round but scored only once. The conventional reading blames bad luck. A reading based on shot locations and shot quality shows the problem lay in the choice of finishing positions. The same table of numbers, two opposite conclusions, and only one of them survives later verification. The problem with data is rarely the data; it is whether the person reading it bothers to trace its source.

The asymmetry of risk screening. This is the least discussed and most expensive part. Severe risks in sport — unpaid wages, signs of match-fixing, undisclosed injuries, sanctions from organisers — share one characteristic: they are silent by default. Nothing automatically surfaces to raise an alarm.

Which means that when a data file makes no mention of unpaid wages, that does not mean the club is paying on time. It only means nobody has checked. The absence of a bad signal is not proof of good health. I got this wrong in many early reports, presenting a clean file because I had found no problems, when in reality I had simply not looked.

With officiating, the same logic applies to VAR. The criterion of "a clear and obvious error" sounds like an objective standard, but the threshold for placing an incident in that category shifts by match, by competition, by whoever is sitting in the room. The space for subjective judgement in VAR is far larger than the way the media describes it, and any analysis that treats VAR decisions as hard facts is building on sand.

The popular belief in sports analytics is that more data produces better conclusions. I do not believe that. What determines quality is not the volume of input, but discipline at the source-verification step.

When I was challenged on a live broadcast during Euro 2026 for arguing that Germany's exit was not down to luck, I was called a data worshipper who disdained the emotion of the game. The rebuttal was partly right, but aimed at the wrong target. People thought I worshipped spreadsheets. In truth I worship the practice of naming sources. If a metric has no collection date, no sample scope, no definition, it is not data — it is a sentence presented in a handsome font.

Another line of criticism deserves to be taken seriously: some things should not be quantified. A player's state of mind after three straight defeats, the weight of a full stand, the breathing rhythm of a squad losing its spirit — none of that fits into a table. At the 2026 World Cup, the champions did not win with the prettiest passages of the tournament. The 2026 World Cup lifted the trophy through tackles nobody remembers. If you only read the scoresheet, you miss the very thing that won it.

So my position is not "data is always right". My position is that data must always be questioned, including when it comes from me.

In a transfer window, the temptation to fill gaps peaks. Readers are drowning in rumour, and every writer wants to be first with an answer. But what readers actually need is not more news; they need a filter that separates claims with evidence from claims with a confident tone.

In a transfer window, noise is always available; signals have to be hunted. Release clauses, wage-bill structures, the movements of agents, the moment a club sells an asset to balance its books — that is where the story lives, and that is where few bother to read.

The question I am carrying into the next analysis cycle is not whether we have gathered enough data. It is what we are failing to screen for. Because what brings an analysis down is rarely wrong data. It is usually data that never existed, retold as though it had.