Nine Dimensions of Esports Analysis and the Lesson of an Empty Dataset

2:47 a.m., Brisbane. I opened the latest report from the analysis pipeline before shutting the machine down. It had all nine sections, all the tables, all the bold headings, all the logical order I had built over the years. Section one asked about the patch. Section two asked about tournament format. Section three asked about the roster. Section five asked about cash flow. And in every answer cell, the same line repeated: “N/A — insufficient information, cannot assess.” Not a few cells. All of them. Article title: N/A. Article source: N/A. Information points: completely empty. Three critical fields still held the verbatim instruction text meant for the upstream processing step, never filled in.

I sat still for about ten minutes. Outside the window, the Brisbane road was as quiet as any Tuesday night. Two options were on the table: delete the file and write something else, or keep it and write about that very moment.

I chose the second, because I have followed this esports industry for twenty-three years, and I had never seen a document say “I don’t know” so plainly.

The file had no typos. It had no formatting errors. It simply had no content. And in an industry where every party is trying to say more, a document that chooses silence becomes the most readable thing of the week.

Context: an industry that earns its living by measuring

Esports is built on an unstated assumption: everything is measurable. Every teamfight can be split into hundreds of frames. Every frame can be labelled. Every label can be counted. That produces an enormous metric system — and that system differs so much between titles that it cannot be translated.

That is why the first problem in any analysis is not “is this team strong or weak” but “which game are we talking about”. League of Legends measures KDA, gold-to-damage conversion, objective control rate. Counter-Strike 2 measures HLTV Rating, opening-kill success rate, impact per round. Liên Quân Mobile and the mobile MOBA group measure placement points, pick-and-ban rate, average game length. Three metric systems, three tournament pyramids, three business logics. Mixing them into one table is wrong from the root.

In that report, the field asking “which game” was also empty. And because it was empty, the other eight dimensions collapsed automatically. The person who wrote the report did the hardest thing in this profession: they stopped.

I have seen the opposite far more often. In 2026, in a press room in Ho Chi Minh City after a group stage of a mobile MOBA event, I heard a host ask a coach about “objective control rate”. The coach laughed. He said he did not know how that number was calculated, because the tournament organiser’s data department produced it and no one published the formula. The room laughed with him. Three days later, that number appeared in the headlines of at least four different outlets.

When the spreadsheet speaks, the stadium has to learn to be quiet. But when the spreadsheet falls silent, this industry usually chooses to talk more.

In Vietnam, esports growth comes with a paradox. The audience grows faster than the number of people who can actually read the data. That gap gets filled with emotional commentary, and emotional commentary is always available. A loss is explained by “spirit”, a win by “mental strength”, and nobody asks further.

In Australia, where I live, the story is slightly different. The market is smaller, but audiences demand sources. International events hosted in Sydney over the years have created a layer of viewers used to reading pre-match analysis. That is an advantage. It is also a pressure: publish a number without a source and the comment section will ask about it immediately.

Nine dimensions, and the cost of skipping any of them

The framework I use has nine layers. I did not invent it. I assembled it from European football journalism, from club analytics rooms, and from my own mistakes.

Dimension one: patch and meta. This is the most underrated layer. A patch does not merely change numbers. It changes the priority order of an entire roster. In League of Legends, when a mid-lane champion loses early-game damage, the team with the best player in that position loses part of an advantage built over thousands of hours. At professional level, assessing a patch needs at least three data groups: champion win rate, pick-and-ban rate, and average game length. Miss one and the conclusion tilts.

There is a technical detail outsiders rarely notice: tournament servers are usually locked to a different version than the public server. Teams practise on one meta and compete on another, weeks apart. In that gap, the team that reads the patch faster wins before the match begins. This is the dimension the report could not reach, because there was not a single version identifier.

Dimension two: tournament system and format. Format determines upset probability. A best-of-three series differs entirely from a single game. An organiser choosing the Swiss system creates a bracket where teams play a very different number of matches, which directly affects who rests more. Vietnam’s League of Legends championship in recent seasons has used a group stage plus a playoff bracket, creating two completely different pressures: the pressure of accumulating points and the pressure of a single elimination game.

Another factor the news cycle ignores is schedule density. A team playing four matches in seven days looks different in the data from a team playing one match in seven days, at the same skill level. Without the calendar, any comment on form is speculation.

Dimension three: team and people. This is where data meets its limit, and where I work most. When analysing a roster, I always check three things first: the average age of the main lineup, the number of matches played together, and who calls the tactics.

Two patterns repeat so reliably that I call them rules. The first is the age cliff: in most competitive titles, reflexes peak around twenty, and the decline is not gradual — it tends to be sudden after a long break. The second is the new-roster honeymoon: a team that swaps two players usually performs very well for six to eight weeks, because opponents have no data on them. After that, once the footage has been dissected, the effect disappears.

And there is a risk group public data never touches: injury. Carpal tunnel syndrome, tendinitis, back pain from thousands of hours in bad posture. I once saw a professional player tape his wrist for an entire season and the organiser never announced it. No data column exists for a player forced to cut his click rate by thirty percent in a teamfight because of pain.

Dimension four: regional landscape. The same country can hold completely different status in two titles. Vietnam sits among the world leaders in some mobile MOBA titles, with permanent slots at international events, but is far more modest in Counter-Strike 2. Japan is strong in traditional fighting games, weaker in MOBA. Korea is nearly the definition of a “tier one region” in a few titles, but not all.

The most common error when discussing regions is using past results to infer current strength. A country that once won a world title may have lost three of its five core players two years later. The trophy cabinet does not update itself.

Dimension five: cash flow. This is the dimension I believe matters most and is also the most concealed. Most professional esports teams carry a salary-to-revenue ratio far beyond any norm in traditional entertainment. Revenue comes from three sources: sponsorship, publisher and organiser distributions, and merchandise. When one source contracts, the team does not cut spending immediately, because player contracts are signed. That lag is when teams disappear.

Having worked with the balance sheets of a few organisations, I learned that the early warning sign is not losing. It is when a team stops announcing a substitute roster, stops recruiting for its academy, and starts posting more sales jobs than coaching jobs.

Dimension six: rules and governance. Esports has a rare structural feature: the publisher is the rule-maker, a commercial stakeholder, and the judge. There is no independent arbitration mechanism equivalent to the Court of Arbitration for Sport in football. That does not mean every decision is wrong, but it does mean every decision lacks a layer of external verification.

Dimension seven: risk profile. I group risk into six categories: competitive, financial, personnel, rules, public opinion, and systemic. The first five attach to specific cases. The sixth sits outside all cases — and it is the one that fired on the night I opened that report.

Dimension eight: public narrative. Every team lives inside a heat cycle: budding, heating up, climax, backlash. The right question is not “what is the public saying” but “which phase of the cycle is it in”. A team overhyped in week three usually pays with a wave of criticism in week nine.

Dimension nine: industry transmission. The chain has three layers: publishers upstream, clubs and streaming platforms in the middle, sponsorship and derivative markets downstream. A small upstream change can take eighteen months to reach downstream. That is why many industry forecasts are right in direction and wrong in timing.

The transfer market and the costs nobody puts on the sheet

One topic I always approach carefully: agents.

In most esports transfers, the published number is the visible part. The submerged part includes brokerage fees, release clauses, image-rights percentages, and performance bonuses that are almost never disclosed. Added together, the true cost of a deal can run thirty to sixty percent above the headline figure.

I am not saying the agency profession is bad. I am saying it has its own incentives, and those do not align with the team’s, the fans’, or journalism’s. When one intermediary’s voice becomes the primary source for an entire market, the value of that information falls.

The concrete consequence is mispricing. A player can be valued above his ability because his agent talks more. Another can be undervalued because nobody speaks for him. In both cases, the performance data never enters the pricing process.

The mobile stage: where data is most neglected

If there is one area where the data shortage is most severe, it is mobile esports. Mobile MOBA titles have enormous player bases in Southeast Asia, with Vietnam among the largest markets. But public data infrastructure is far thinner than for PC titles. There is no frame-addressable footage archive. There is no public API for advanced metrics. Most analysis is written from the writer’s direct impression.

That creates a paradox: the largest audience sits in the area with the least evidence. And when evidence is scarce, power shifts to the storyteller.

The contrarian angle: the problem is not a shortage of data

I have spent most of my career convincing people that data matters. That night, I realised I had the emphasis wrong for years.

The problem with esports analysis is not a shortage of data. It is a surplus of conclusions. Thousands of articles are produced each day from a roughly constant amount of information, and article volume is not proportional to event volume. When articles vastly outnumber events, the excess can only come from two sources: reinterpreting what exists, or inventing what does not.

Here is the point I want to make clearly, because it runs against my own instinct: an analysis returning “insufficient information” across all nine dimensions is not a professional failure. It is a success at a lower layer — source control. It stopped a fake product that looked very real.

Correlation is not causation, and fluency of prose is not evidence of authenticity. A paragraph can be smooth, numeric, full of names, and entirely untrue. In the past eighteen months I have read analyses where figures appeared to two decimal places with no source at all. Readers have no way to tell.

This industry has an implicit bias: silence is read as weakness. An expert who says “I don’t have enough data yet” is rated lower than one who says “I am certain”. In sport, that produces a concrete outcome: the certain ones appear on media more often, and over time the public builds its image of expertise from the least accurate people.

At 39, I learned that data also hurts when it is distorted. A metric placed in the wrong spot does not stay still. It spreads. Into articles, into videos, into audience comments, and eventually back as pressure on the players themselves.

There is a further consequence I rarely state publicly. Automated content systems are driving the cost of producing fake analysis toward zero. That means the economic reward no longer sits in writing but in sourcing. Newsrooms that move first will build an advantage that cannot be copied: they own footage, data contracts, and people who sit and watch. The rest will compete on speed, and speed does not require truth.

This is where I disagree with most colleagues. Many believe the wave of automated content will destroy analysis. I believe it will filter it. It will destroy the lowest part of the market — the part that rewrites press releases and adds three adjectives — and raise the value of the highest part, the part requiring hours in front of a single clip.

Why I still watch with human eyes

There is a kind of information a spreadsheet can never carry: the silence between two actions.

I remember an evening in Sydney, at an international event, sitting in row eleven. On the big screen, round-by-round metrics displayed perfectly. But what caught me was a player putting his mouse down during a break, rotating his wrist three times in the same direction, then picking it up. Three times, same direction. Afterwards he changed how he entered fights.

No data column records that. But it is data — just data in a form spreadsheets have not learned to read.

That is also why I refuse to abandon my old way of writing. I still open the advanced metric sheets and the positional tracking maps, but I always spend the first ten minutes looking without writing anything. Every number has a story; my job is not to ruin it. But to find the story, sometimes I have to switch the numbers off first.

The long shot in memory always hits the top corner; in the spreadsheet it flies straight at the keeper. Both are true. It is just that the first cannot be reproduced, and the second can.

In a match I rewatched four times, the only thing that changed across those four viewings was not the number. It was my understanding of a person. First I saw a mistake. Second, a rushed decision. Third, a man who had just lost the teammate he trusted most. Fourth, I saw nothing at all, because I understood.

My spreadsheet stayed identical across all four. Only I changed.

What I took from that night

I did not delete the report. I renamed it to something that reminds me, and I put it in the first folder I open each morning.

Three things I consider worth doing, and I am doing them.

First, I set a hard condition for any analysis pipeline I am involved in: if a core information field is empty, the pipeline must stop and raise a loud error. A system that halts is a system working correctly.

Second, I write absolute dates on every citation. “This week” and “recently” are not dates. In an industry where a season turns over every six weeks, one word can be off by a year.

Third, I keep empty analyses as specimens. They are evidence of something this industry says very little about: the limits of knowledge are part of knowledge.

Esports is where European football stood in the early 2000s, when clubs began hiring data analysts and were mocked for it. The people mocked that year were the ones who turned out right. In the A-League, I was once called a rebel simply for bringing a laptop into the press room. Twenty years later, every club has a data room.

So when someone asks me which team is stronger this week, I answer with a question first: how many hours of footage will you give me, and do you have the patch identifier.

If the answer is no, I will tell them exactly what that report told me, and I will not be ashamed of it.

Because an analysis with no content, in an industry full of wrong content, is a rare form of honesty. And in a season still stretching ahead, that honesty will be the last thing standing when the spreadsheet finally speaks.

Reference note: Stage-2 deep professional analysis report, esports domain, internal document with no recorded publication date, supplied by the operator of the analysis pipeline; tournament structure and performance metrics cross-checked against the VuaBong.vn database.