Data Integrity in Esports Analysis: Lessons From a Report With No Subject

On the morning of August 13, 2026, a nine-part analysis file landed on my screen. The tables were drawn neatly. Every section had a cell. Every cell had text. And almost all of that text was the same repeated marker: insufficient information.

The first instinct of someone who has spent eighteen years reading club balance sheets is to look for the broken part. The second instinct — the far more dangerous one — is to want to fill the blanks. The brain of a budget professional keeps insisting that an empty cell is an error to be corrected, not a fact to be published.

I have made exactly that kind of error before. And its cost never showed up on the payroll sheet. It showed up in the league table.

The file I received contained nine analytical dimensions: patch and meta, tournament system and format, team and players, regional landscape, club finance, rules and governance, risk profile, public narrative and expectation, and finally industry transmission. Nine dimensions, nine tables, and not one number in any of them.

The professional conclusion of that review was not an esports judgment. It was a diagnosis of a data pipeline. That sounds unglamorous. But in an industry where thousands of analyses are published every week, a pipeline diagnosis is the single most valuable piece of information an analyst can produce.

A report that is complete in form and empty in substance is a more dangerous product than one that is openly blank, because a reader cannot tell the two apart at a glance.

---

Context: the two-stage analysis pipeline

In the kind of deep analysis work I take part in, the process is split into two clearly separated stages. Stage one performs deconstruction: it reads the source text and extracts the title, the source, the article type, a one-sentence summary, the author's stance, the article's purpose, a list of information points, the entities named, the time sensitivity, and the source quality. Stage two is where domain interpretation happens.

That split is not decoration. It exists because esports has one deeply inconvenient property: information changes faster than it can be verified. A patch can invert an entire tactical meta within fourteen days. A transfer window can turn a championship roster into a mid-table roster within three weeks. A publisher rule change can erase a tournament slot a club already paid to obtain.

In that environment, if the extraction step fails while the interpretation step still runs, the output is not a weak analysis. The output is an analysis of the wrong subject delivered in a confident tone.

That is why I am writing this. Not to recount a technical incident, but to name three failure mechanisms I believe operate quietly inside a great many esports analyses published every day, including in Vietnam.

Failure mechanism one: silent subject substitution

This is the most dangerous mechanism, and the hardest to self-detect.

Subject substitution happens when the input data lacks a core identifier — a game title, a patch number, a team name, a tournament — and the analyst fills that blank by inference from surrounding context. A keyword in the task title. A familiar word in the opening paragraph. A name recently read somewhere. One plausible fragment is enough for the analytical machine to keep running, and it will run very smoothly.

The problem is this: a patch analysis written about the wrong game does not read like a wrong analysis. It reads like a normal one. Correct structure, correct terminology, numbers that look reasonable. Only the subject is wrong, and sometimes only by one word.

I know that feeling from the other side of the table.

In 2026, when I was twenty-five and had just started doing financial analysis for Beijing Guoan, I proposed spending twelve million euros on the midfielder Jonathan Viera. My basis was his key pass and expected assist numbers in La Liga. The numbers were beautiful. The trend line pointed up. I built a valuation model, presented it in a meeting, and got approval.

I ignored a single variable: his capacity to adapt to Chinese football. Not the quality of the league. The cultural distance, the tempo of movement, the way referees call fouls, the way defenders treat a high-priced foreign player.

Six months later, Viera's form collapsed. The board sold him for eight million euros. The four million euro loss sat plainly on the books, and in a closed meeting the head coach named me directly: "Numbers cannot replace direct observation."

I do not tell this story to burnish myself with a failure. I tell it because it is the same class of error: I substituted a missing subject — the player's capacity to adapt to a new league — with a plausible assumption, and I wrote forward from that assumption without verification.

The market does not forgive, it only records — and I paid for it with the 2026-18 season.

Since then I have set a personal rule that cannot be broken: every number must be cross-checked against a minimum of three real match contexts. I never publish a single indicator without its evaluation conditions attached.

Based on my experience watching matches, a successful cross count means nothing unless you state whether the player crossed against a high defensive line or a deep one, at what score state, in the tenth minute or the eighty-fifth, and whether the opposing right winger was suspended.

Three contexts. That is the minimum threshold. Not because three is sacred. Because by the third context, the denominator starts pushing back against the story you wanted to tell.

---

Failure mechanism two: treating null values as a shame

In the second-stage analysis I was diagnosing, the correct handling of an empty input is to state it plainly: insufficient information, cannot be assessed. Record it clearly. Do not infer.

That sounds simple. But it is the kind of sentence very few analysts are willing to write, because it feels like a confession of professional failure. You are paid to deliver judgments. You just filed a ten-page document consisting entirely of "cannot be assessed."

The truth is the opposite. In valuation work, the sentence "insufficient information" is a conclusion with the same value as "expensive" or "cheap." It protects decision-makers from wiring money into a model built out of air.

There is one technical distinction I want to stress, because it is often collapsed into a single category. Degraded extraction is the case where part of the data is right and part is wrong. A total null is the case where there is nothing at all.

The second sounds worse. It is in fact far easier to diagnose, because the error shows up in every cell at once. The first is the real nightmare, because the mistakes hide inside cells that look correct. A name extracted correctly, a number extracted off by a decimal, a timestamp reversed. Nobody catches that without an origin to check against.

That is why I always tell young analysts the same thing: the most dangerous error is not in the empty cell. It is in the cell that looks reasonable enough that you skip the check.

---

Failure mechanism three: screening asymmetry

This is the part I most want Vietnamese esports readers to remember.

In this industry, most severe risks are silent risks. They do not surface on their own. They surface only when someone actively goes looking.

Unpaid wages are one example. A club that stops paying players does not issue a press release. The news leaks through a two a.m. livestream, through a status update deleted thirty minutes later, through a contact inside the management team.

Match-fixing is a second example. No violation announces itself.

A star player's injury is a third. The medical staff know. The coaching staff know. The fans find out after the team has lost two matches.

What these three share is that they belong to the category of risks that appear only under active screening. Without screening, they do not appear. And when they do not appear in the data, readers easily default to assuming they do not exist.

That is the single most serious logical error in the entire sports analysis industry.

An empty financial column is not a healthy balance sheet. It is an unchecked balance sheet. Those are two different states, and the distance between them can be three months of a roster's salaries.

I learned this lesson in a season with no spectators.

In March 2026, when every league in China was suspended because of COVID-19, I was working at Shanghai SIPG in a mid-level role. Ticket revenue was zero. Matchday revenue was zero. But the labour contracts were still fully in force.

I immediately proposed a plan to cut thirty-five percent of non-essential operating costs. Cancel the private bus contract. Renegotiate the data analysis fee with Opta. Review every small line item, including the cost of printing match documents.

That plan saved the club 2.3 million RMB in the second quarter. That money was enough to keep two Brazilian assistant coaches who had originally been placed on the exit list. For two weeks I worked eighteen hours a day, building an emergency plan down to the individual cost line.

When the stands are empty, I can hear the voice of every single unit of budget.

The lesson I drew, and still apply, is this: every article I write about a financial crisis follows a fixed three-layer frame — cash flow, liquidity, recovery capacity. And every article contains at least one concrete budget table.

Apply that frame to the empty report I was diagnosing. No revenue line, no salary line, no capital injection line. You cannot build a cash-flow frame from three blank rows. You can only state that the frame has not been built.

That is not a weak judgment. It is an accurate one.

---

The illusion of framework completeness

There is one more risk nobody in the industry states clearly enough.

A nine-part document, with a table in every part, a row in every table, a bold label on every row, will be read as a document with weight. Table density creates a feeling of authority. That is the reader's reflex, and in most cases it is a reasonable one.

But when the substance is zero, that same density becomes camouflage.

I re-checked all nine tables in the file I received. Patch impact table: cannot be assessed, in every cell. Format structure table: cannot be assessed. Roster assessment table: cannot be assessed. Regional strength comparison: cannot be assessed. Financial structure table: cannot be assessed. Compliance checklist: cannot be assessed. Risk matrix: cannot be enumerated. Expectation gap table: cannot be assessed. Industry transmission map: cannot be assessed.

Nine tables. Not one fact.

A reader skimming sees a professional document. A reader reading carefully sees a blank blueprint.

For an industry where most readers encounter information through screenshots of tables on social media, the gap between those two kinds of reading is the gap between information and noise.

---

Every analysis depends on a core identifier

There is a technical principle I consider the most important in the whole workflow: patch analysis, roster analysis, and regional analysis all depend on the game title. Without a title, those three dimensions cannot run.

The reason is concrete. The same region can be a tier-one region in one title and a wildcard region in another. The same roster can be strong in a push meta and weak in a vision-control meta. The same transfer move can be an upgrade in one league and a mistake in another, purely because the import rules differ.

Assigning a regional tier by intuition corrupts every downstream conclusion, in a way nobody detects, because every subsequent step is reasonable on a false foundation.

Tournament tier carries similar weight. A world championship, a regional league, and a third-party invitational have different upset rates, different preparation windows, and different governance risk. Lumping them into one category is the fastest way to produce an analysis that is technically correct and predictively meaningless.

---

Valuation: when a number can actually be used

The one part of that empty document I can speak about with high confidence is the financial part, simply because I know exactly what is missing.

To make a judgment on a transfer fee, you need a minimum of two things: a specific fee and a benchmark. The benchmark can be the fee for a player of the same position, same age, same league, in the same transfer window. Without those two, every judgment of "expensive" or "reasonable" is a feeling dressed up in terminology.

I once ran that process correctly, and the result made me rethink how I value players.

At Euro 2026, I was assigned to write a fast financial brief for a tactical analysis outlet. I noticed that Italy's left wing-back Leonardo Spinazzola had ten successful crosses into the box in the first four matches. Wingers of comparable level averaged roughly five over the same number of games.

I built a transfer valuation formula based on expected threat from the left flank and test-applied it to five top Premier League clubs. The brief was shared more than two thousand times on Weibo, and a player agent contacted me proposing a partnership to track the market.

Spinazzola does not take free kicks; he stamped a new valuation rule.

But the most important part of that brief was not the formula. It was the note attached to it: sample size of four matches, limited to one short tournament, applicable only to wing-backs inside a system with a left-leaning holding midfielder. Without that note, the formula would have become an unfounded prophecy transmitted two thousand times.

And I have also been wrong in the opposite direction, on a far bigger deal.

In January 2026, when Julián Álvarez was still at River Plate, an acquaintance inside the City Football Group asked me: "Can you believe twenty-one million euros?" I reviewed six months of his numbers: fourteen goals and six assists in Argentina, but a very low true tackle figure. I concluded the risk was high, because form in South America says little about adaptability in the Premier League.

Manchester City signed him. In the 2026-23 season, Álvarez scored seventeen Premier League goals. I was wrong.

But that mistake forced me to rebuild my evaluation method. I added weight to two variables my old data set could not measure: live-ball situations and the ability to create space for team-mates. I also added a mandatory section to my process called "Why the data can lie to you," with Álvarez as the standard example. At the end of every piece, I recommend readers verify using at least two independent data sources.

I learned valuation from one mistake, and never needed a second lesson.

---

Governance and competitive integrity: a risk class that cannot default to clean

Of the nine dimensions in that document, the one that held my attention was rules and governance.

This is the highest-severity risk class in the entire industry. A match-fixing allegation can end a player's career, dissolve a team, and cost a club a slot it paid to obtain. A publisher rule change can erase a working business model.

The problem is that these risks do not surface on their own. An empty input cannot confirm they exist, and cannot confirm they are absent. Their correct status in the absence of data is unscreened.

This is where I see many esports analyses in Vietnam and the wider region get it wrong. When there is no news, they assume nothing is happening. In verification practice, no news means not yet checked.

The same logic applies to governance disputes with publishers: mid-season rule changes, revenue-share conflicts, allegations of double standards in sanctions. Without a named publisher, title, and league, there is no way to assess any of it.

---

Public narrative: expectation needs a denominator

Every analysis ends with the gap between market expectation and objective reality.

To judge whether a team is overhyped, you need two terms: an expectation term and a fundamentals term. The ratio of social-media heat to competitive fundamentals can only be computed when both terms exist. Without one, you do not have a ratio. You have a number divided by empty space.

With an empty document, both terms are missing. The correct conclusion is that the degree of overhype cannot be assessed.

That is the kind of conclusion sports media usually skips, because it generates no headline. But it protects readers from believing in an emotional cycle built out of nothing.

---

The contrarian angle: the correct professional act is to refuse to write

Here I want to state plainly what I believe is the most important point in this entire article.

The entire sports analysis industry pays for volume. More pieces, more threads, more comments, more engagement. The incentive structure rewards people who speak a lot and speak fast. In that structure, refusing to write is treated as laziness or incompetence.

I think it is the reverse.

In risk analysis work, the most correct professional act is often to refuse to deliver a conclusion when the data is insufficient. You are not rewarded for it. You are often penalised for it. But it is the only boundary between an analyst and a machine that generates plausible text.

And there is a practical reason I hold this line.

False information travels faster than corrections. A wrong claim about a patch, a wrong claim about a roster, an invented transfer figure, will be shared thousands of times within hours. The correction, if it comes, will be shared a few hundred times within days. That ratio is not unfair. It is the structure of social media, and nothing will change it.

Which means the cost of one filled-in blank does not stop at that article. It spreads into other people's decisions: a team picks the wrong player, a sponsor puts money in the wrong place, a fan places trust in the wrong person.

The second point I want to make is somewhat paradoxical.

The higher the degree of automation, the more dangerous the illusion of completeness becomes. A workflow can generate ten pages of tables in thirty seconds. But it has no mechanism to know it is talking about a subject that does not exist, unless someone builds that mechanism in. And that mechanism, in practice, is a single line: if a required field is empty, stop.

That sounds too simple to believe. But most failures in this industry do not come from complex algorithms. They come from the absence of a stop command.

Third: probabilistic language needs a specific referent. Words like "may," "tends to," "high probability" only mean something when attached to a defined subject. When the subject does not exist, they are not caution. They are noise wearing an expert tone.

Fourth, and this is the technical detail I want sports editors to remember: "data pending verification" and "no data exists to verify" are two different states. The first assumes a number exists and needs checking. The second states that no number exists at all. Collapsing the two creates a false belief that a verification process is underway, when in fact there is nothing to verify.

---

Applied to the Vietnamese context

I follow Vietnamese esports from an operational angle, not a fan angle. And I see three distinctive features in the data problem here.

First, the information ecosystem is dominated by unsourced transfer rumours. A name is mentioned on a livestream, and within twenty-four hours it becomes a deal under negotiation, then a completed deal, then a collapsed deal. None of the three versions has anyone putting their name to it.

Second, performance data at domestic league level is often not published in full, so any quantitative analysis must rest on small samples and must state that it is a small sample. When sample size is unstated, readers default to assuming it is larger than it is.

Third, time pressure in Vietnamese sports media is extremely high. A match ends at eleven p.m. and the analysis must be live within an hour. In that window nobody has time to check three contexts. But that does not license skipping verification. It means choosing to write less, not to write more carelessly.

A tight budget does not produce poverty; it produces sharpness. That principle applies to budgets and to information capacity alike.

---

The correct process when the input is empty

If an editor or analyst reading this recognises a similar hole in their own workflow, here are the concrete steps I propose.

Check whether the source text was actually retrieved. HTTP status, access rights, paywall, JavaScript-rendered page, character encoding errors. Four of those five failure types do not surface as error messages. They surface as an empty field.

Re-run the extraction step and confirm the information point list is non-empty before triggering interpretation. That is a technical gate, not a moral rule. It is simply a necessary condition.

If the source text genuinely contains no extractable entities, the correct output at the interpretation step is a short notice that the piece falls outside analytical scope. Not a nine-dimension report.

And when an analysis depends on the game title, establishing that title must be the first step, before any other table. I have seen too many documents that begin with meta analysis and end with a footnote saying the game title is undetermined.

---

A forward-looking thought

The nine-part document I received on the morning of August 13, 2026 contained not one fact about esports. But it taught me something eighteen years of reading balance sheets had not fully taught me.

In this industry, the value of an analyst will increasingly be measured by the number of conclusions they refuse to deliver, not the number they publish. Because text-generation tools have become cheap and fast, the scarce resource left is not the ability to write. It is the ability to know when to stop.

The next generation of analysts will not win by writing more. They will win by building a workflow with a gate in it, and by holding that gate when the pressure to publish arrives.

As for me, I keep the old rules. Every number comes with three contexts. Every conclusion comes with its conditions of application. And when there is no subject, the first thing to do is not to write — it is to find out where the subject disappeared.