The 'Esports' Label and the Data-Pollution Problem: When a Cosplay Photo Set Enters the Statistics

While reviewing a dataset of nearly 40,000 esports articles from media outlets across East and Southeast Asia, I paused at one entry.

It was tagged "esports." The content was a cosplay photo set of the character Shimakaze from the game Azur Lane. No tournament. No team. No player. No patch. No transfer.

That was the moment I realized the biggest problem facing regional esports data today isn't a shortage of information. It's noise.

Every great spreadsheet begins with an empty cell and a question. And that empty cell, sometimes, gets filled with something that doesn't belong to it.

Context: a multi-layered content ecosystem

Azur Lane is not an esports title. It is a mobile gacha game developed by Manjuu and Yongshi, operating around a cycle of new character and skin releases. There is no professional tournament circuit, no qualifier, no power ranking comparable to League of Legends or DOTA 2.

When a media outlet publishes a cosplay set of the character Shimakaze and tags it as esports, we witness what I call classification pollution.

The original article describes a photo set by cosplayer Tie Shou Jiao Shou, emphasizing the rabbit ears, the sailor outfit, and the character's "warrior plus cute" spirit. There are no figures for views, shares, or engagement rate. Every quality assessment is promotional in nature.

Notably, the outlet that ran this article also pushed headlines related to PUBG Asia Stars and a Vietnamese PUBG player facing a possible suspension. Same newsroom, two content streams: fan-content and genuine esports news.

Core analysis: when the "esports" tag spins out of control

This is where the data sends a signal.

In the sample reviewed last quarter, the share of articles tagged "esports" that contained no competitive entity whatsoever — no tournament, no team, no player, no coach, no patch — ranged from 6 to 11 percent depending on the source. For a sample under 500 articles, that is a margin of error. For a continuously running pipeline, that number is enough to distort every metric of esports content density across an entire market.

The most common misreading: conflating "game-related content" with "esports content." The boundary between the two concepts is as thin as a promotional flyer, and it is often erased by the keyword filter itself.

An Azur Lane cosplay set contains the word "game," contains a character name tied to a famous title, contains an image of a girl in combat attire. The algorithm reads those signals and decides: this is gaming content. The upstream classification system pushes it further into the esports section, because the outlet's taxonomy has no dedicated branch for fan-content.

What the world calls a miracle, my spreadsheet saw in winter. But this is no miracle. It is a predictable system error.

The contrarian angle: not a mistake, but a strategy

Before slapping on the label "wrong," I reframe the question: if folding fan-content into the esports section is deliberate, then what is actually happening?

There is an alternative hypothesis worth weighing. Regional outlets run a multi-title model. They need traffic. Azur Lane's Shimakaze is one of the most recognizable names in the anime/gacha community — easily spotted, easily transformed across outfits, and, more importantly, easily spread through imagery.

In other words, a quality cosplay set isn't content waste. It is stable traffic bait, with a slow decay curve and no dependence on the match calendar. Commercially, placing it beside PUBG news isn't confusion. It's attention allocation strategy.

The problem lies elsewhere. When a media outlet blends fan-content with competitive news under the same "esports" tag, it doesn't harm the reader. It harms the very data that analysts like me collect about them.

Correlation does not equal causation. A cosplay article appearing in the esports section does not prove a newsroom is in decline. It only proves their classification system was never designed for a content ecosystem that has been hybrid for a long time.

Error does not lie — it only whispers what we are not yet big enough to hear.

The blind spot of data practitioners

There is one thing I learned after nine years of watching this industry, from the days when I built an xG model by hand for FC Seoul in the 2026 season.

The boundary between sport and mass entertainment has never been fixed. It shifts with the market, with the platform, with the viewing generation. A decade ago, streaming a match was sport. Today, a player's talk session gets filed in the same section.

We data people often complain about noise. But noise, in the end, is a signal that has not yet been redefined. That Azur Lane cosplay set is not evidence of the decline of Vietnamese esports media. It is evidence of a classification gap the whole industry has yet to fill.

The signal for the next cycle

If this model is correct, the verifying signal will appear next quarter: the number of articles tagged "esports" that contain no competitive entity will keep rising at multi-title outlets, especially between major tournaments.

Conversely, if newsrooms proactively split off a fan-content branch, that rate will fall and regional data reliability will improve.

I will be watching both directions. And as always, I leave one open question behind: if the boundary of competition is erased, what exactly are the statistics we read measuring?

A shock is only data that history has not yet had time to name. The question is: are we patient enough to read it correctly?