One Spreadsheet, Twenty-Eight Lines of Leaked News, and Six That Have a Source

One Spreadsheet, Twenty-Eight Lines, and Six That Have a Source

At 02:47 in the morning, I opened a blank spreadsheet and started counting.

On my screen was a freshly published story from a Vietnamese games outlet, bylined Tuấn Hưng, about a remake being speculated over within the survival-horror fan community. I did not read it to learn its content. I read it to classify it. Every claim in the piece I separated into its own cell and tagged at one of three levels: confirmed by the publisher, inferred from real facts, or unsourced rumour.

When I reached the twenty-eighth line, I stopped. Six lines traced back to an official source. The remaining twenty-two floated free.

That structure has recurred too many times over the past decade for me to treat it as an exception. I treat it as a pattern. The more detailed the story, the more credible it feels. Feeling is not evidence. Across twenty-three years of watching sports and esports markets, I have learned one costly lesson: the specificity of a piece of information does not correlate with its veracity, and in many cases high specificity is itself a marker of a story built to optimise for pageviews.

This is why I am writing this piece. Not to discuss a particular game, but to reconstruct a verification process that anyone tracking sports, transfer, or esports news should have at hand.

Context: why I classify instead of judging

My method grew out of a professional shock.

In 2026 I was a mid-level staffer at a sports channel. I was assigned a pre-match analysis for a World Cup qualifier, and I built my argument on expected goals and progressive passes. My conclusion was that the national team should play possession football. The match ended goalless, and the team only advanced thanks to fortune in the final round. The next day a male colleague said in front of the whole newsroom that I knew only how to cling to statistics without understanding football.

I did not argue. I went home, downloaded all thirty-eight qualifiers across five confederations, re-split them into smaller datasets, and cross-checked every metric against the footage.

The lesson lay elsewhere. That mistake taught me that data never lies, only the reading of it is wrong. I had used a correct metric to answer a wrong question. Expected goals measures the quality of chances, not the capacity to control a match. I put two different things on the same scale and drew a conclusion.

Since then my process has carried one mandatory step: before evaluating any claim, I must establish what question it is answering, and whether the data I hold actually measures that question. The step sounds obvious. It is not. Most wrong conclusions in sports and esports analysis do not come from bad data but from using good data to answer the wrong question.

The story I was analysing that night is a perfect example of this error — at a higher level.

The core: reconstructing the evidence chain

Let us start with what can be anchored.

The publisher confirms three things. First, this is a remake built on the company's in-house engine, in third person. Second, the source material is a title that launched in 2026 on the Dreamcast. Third, the team behind it is the group that made the two most recent remakes in the franchise. These three lines have roots. These three lines are traceable.

The rest of the story is many times longer, and all of it sits in the grey zone.

The list of speculation includes: a semi-linear structure granting players more freedom to explore; a crafting bench; weather and injury states affecting character appearance; the return of a signature enemy type; an expanded romantic thread between two leads; the appearance of a beloved character; and an expanded confrontation between two rival characters.

This is where I want to linger longest, because it contains a subtle trap.

In transfer analysis I have met the same pattern. The more detailed a deal is described — fee, contract length, add-ons, shirt number — the more readily the community treats it as done. But when I cross-check, deals described in excessive detail tend to carry no higher a hit rate, and sometimes a lower one, than deals mentioned in a single line.

The reason is technical. A genuine source usually reaches only part of the matter: one person knows the salary, another knows the duration, a third knows the release clause. Nobody holds the whole. When a story presents every detail fitting together so perfectly, the likelier explanation is that those details were built from a ready-made frame and filled in by inference, rather than gathered from multiple independent sources.

I do not trust intuition; I trust numbers that speak once they are asked the right question. And the right question here is: how many independent sources does this information come from, and can those sources corroborate one another?

Apply that question to the story under analysis and the answer is: unclear. The story itself admits that most of its content sits at the level of unconfirmed leaks. That admission matters greatly, and readers usually skip it because it is buried near the end, after they have consumed the exciting part.

The classification problem: a story outside the field it was filed under

Here I must state plainly what I consider the single most important finding of the entire exercise.

This story, by the nature of its content, does not belong to the esports field. It has no teams, no players, no tournaments, no balance patches, no competitive meta, no clubs, no governance questions. It is a games-industry story — specifically a rumour-aggregation report about a single-player title. The only entities that could be called people within it are fictional characters in a plot.

This confusion is not trivial. It is a systemic error.

I have seen it many times in the trade: a piece of writing is mislabelled at the input stage, and the entire downstream processing chain automatically generates analyses that do not exist. When someone is forced to apply a competitive-meta framework to a single-player title, they will invent things like champions who benefit or win rates for a product that has no win rate. That is how false data is generated out of a process that looks professional.

I have encountered a similar case at a lower level. In 2026, scanning data from dozens of European domestic leagues for prospective centre-backs, I found a Swedish player of Ethiopian origin at an Italian club. He recorded nearly three successful tackles per match, but what drew my attention more was that his successful line-breaking passes exceeded two-thirds of his appearances. That indicated an ability to launch attacks, not merely to defend. I wrote a comparison piece setting him against a leading centre-back of the same age.

When I proposed that scouts consider him, they declined for lack of a direct source. Four months later a bigger Italian club signed him and he became a pillar of a European title-winning campaign. My data was right. But being right was not enough to beat a system that trusts only direct observation.

Both stories — the scout rejecting data, and the editor misfiling a domain — point to the same weakness: people handle categories better than they handle evidence. Once a label is attached, they stop questioning the label.

Sidebar noise and the decoy trap

There is one detail I want to record here, because it illustrates clearly how noise enters analysis.

The outlet carrying this story runs a related-headlines section beside the article. In it appear the name of a leading esports team, a famous player, a regional tournament, and a transfer between two organisations.

Those items are navigational furniture. They are not part of the article's content. But when a text-processing system sweeps the page and picks up character strings, it may sweep this furniture into its input, and from there generate an esports analysis that never existed.

This is the mechanism behind most classification errors in the information industry today. Content and display context are blended. An advert beside an article becomes an event in a report. A related headline becomes a subject.

The prevention is simple but demands discipline: always establish the boundary between main content and accompanying display. In this case, strip away the furniture and the article's true subject contracts to exactly one thing — a single-player game being remade. Every analysis of a leading team, of a player, of a regional tournament is a product of noise, not data.

I have made the opposite error myself. In 2026, when the pandemic forced a domestic league to postpone indefinitely, I analysed one club's data to forecast its survival chances. I found the squad's average running distance was third-lowest in the league, and the rate of tactical fouls in its own half had risen sharply. I wrote a critique of the head coach's tactics. The newsroom refused to publish it, citing a sensitive moment.

I kept the piece. Three weeks later the coach was replaced. The team switched formations but could not save its season.

The cancelled Seoul derby of 2026 is the test case for every predictive algorithm. It taught me that an out-of-model event can invalidate an entire chain of reasoning even when every internal step is correct. In the case of a leak story, the out-of-model event is a publisher issuing an official statement. A single press release can wipe out twenty-two lines of speculation in my spreadsheet.

The contrarian angle: specificity is not credibility

Most readers believe a tacit rule: the more specific the information, the more trustworthy. I hold that with pre-announcement leak news, the rule inverts.

The reason lies in the incentive to produce the information. A genuine internal source has limited incentive: to reveal just enough to test reaction, or just enough to trade for some advantage, without exposing themselves to legal risk. Genuine sources therefore emit fragmentary, unanchored information. A source built to attract pageviews, by contrast, has a very clear incentive: to shape the story to match what fans want, and the more specific the better, to create the sensation of certainty.

In other words, specificity is an index of reader demand, not of source quality.

This produces a measurable long-term consequence: an expectation bubble. When the community absorbs a set of attractive details, it begins building expectations on them. When the actual product arrives without those details, the gap between expectation and reality produces a wave of disappointment, usually out of proportion to the product's real quality.

I have watched this cycle many times in sport. A club described as about to complete a blockbuster signing never completes it, and supporters turn to blame the board for a plan that never existed. The expectation was created by rumour, but the anger is poured onto reality.

The remake story sits precisely inside this template. It sets out changes designed to please long-time fans: a supporting character given an image rebuild, a romantic thread expanded, a classic confrontation made larger. These are the details with the highest appeal, and also the hardest to verify.

There is a further subtlety worth noting. Giving a character an image rebuild is not an action taken by a publisher in the real world; it is the writer's interpretation of a hypothetical narrative decision. Put another way, it is opinion presented in the form of intent. That boundary is often erased in aggregation stories, and readers absorb it as confirmed fact.

If I had to pick one line to warn readers about this kind of news, it would be this: information without a root does not become more credible through repetition. It merely becomes more familiar.

The industry angle: who benefits from a leak

At the market level, a pre-announcement leak is not a random event. It has a structure of interest.

For the publisher, a controlled leak generates free publicity. It measures community interest before official marketing spend. It also allows testing reactions to controversial changes without commitment.

For the media ecosystem, leak news is high-value raw material: low legal risk if merely cited, but large pageview potential.

For fans, leak news offers the sensation of entering a story early.

These three interests resonate, and the result is a stream of unverified information with a very strong motive to exist and spread. This is why I do not ask first whether the news is true. I ask first who benefits from it.

In transfer analysis that question helps me a great deal. An agent benefits when a deal is rumoured because it applies upward pressure on price. A club benefits when a rumour diverts attention from internal problems. Once I identify the motive, I filter out most of the noise before spending time verifying every detail.

The blind spot of a data model

One thing I always stress in discussions with colleagues: every data model has a blind spot, and the blind spot usually sits at the boundary between what is measured and what is not.

In this case my model measures the number of sources, the degree of confirmation, and the presentational structure. It does not measure something important: the existence of a real build in development that I have never accessed. I can classify a story as unverified, but I cannot assert that its content is false.

This is the ethical boundary of the analytical trade. Insufficient verification is not the same as denial. An unverifiable leak can still be true. The problem is that it does not yet meet the threshold to serve as the basis for any decision — whether a purchase, an investment, or a wager.

Every season is a ritual, and the analyst is merely the one who records the omens. This season's ritual is a pre-announcement leak cycle, and my task is to record it faithfully, with a confidence level attached to each line.

Signals to track in the next cycle

For a story in a pre-announcement holding pattern, its value decays fast. I set three signals to track, each with a clear trigger condition.

First, an official announcement from the publisher. Any new confirmed detail resets the whole narrative, confirming or refuting the speculation.

Second, confirmation of story changes. If the publisher speaks to the rumoured character threads, the speculation converts into fact, or triggers a fierce reaction wave.

Third, a release window. Once a date exists, all leaked content is re-anchored on a real timeline.

To me these three signals matter more than all twenty-two speculative lines combined. They are the only anchors usable for the next analysis.

I closed the spreadsheet near four in the morning. Six lines rooted. Twenty-two pending. That ratio, after twenty-three years, remains the occupational average. Readers want an answer now. Data gives me only a condition of waiting.

And perhaps that is the hardest part of reading news in an age when everything is presented as already settled.