When the Data Sheet Is Empty: Notes on Integrity in Esports Analysis

I opened the file at 6:47 a.m., Jakarta time.

Outside the window, the first rains of the season were falling over the Tebet district, and the sound of motorbikes on Basuki Rahmat was not yet thick enough to break the silence of the early morning. I had been sitting in front of the screen for twenty minutes, coffee still hot in the cup, and on the screen was a file named "Stage-2_Final_v4" — thirteen pages long, with all the section headers, all the tables, all the checkboxes.

But every cell was empty.

Not empty in the sense of data that had not yet been poured in. Empty in the sense that the structure existed but the substance did not. Like a stadium whose stands are built, whose ticket gates are installed, whose VAR room is ready, whose lighting system is wired — and then everyone forgets that no one ever scheduled a match.

The original article title: none. Source: none. Article type: unclassified. One-sentence summary: blank. Author stance: undetermined. Article purpose: undetermined. Information points: an empty list. Entities involved: "identify from the information points above" — while the information points above did not exist.

Only one field had been filled: the domain label — "esports."

And I knew, the moment I read that line, that I was facing the worst temptation in this profession: the temptation to write a report that sounds complete about an article I had never read.

Data never lies — only the way we listen is wrong. But here, there was not even data to listen to. The problem was not that the data was wrong. The problem was that a frame had been built in advance, and that frame was exerting pressure on me to fill it with anything at all.

That is the subject of this piece. Not a match. Not a team. But an occupational disease: the disease of reports that look complete but contain not a single truth inside.

---

Context: two tiers of analysis and the gap between them

To help readers understand what I am talking about, I need to sketch how a professional analysis pipeline operates.

In my work — data consultant for football clubs, and writing on esports for the Indonesian market — everything usually runs in two tiers. Tier one (Stage-1) is the extraction tier: read the source article and pull out the title, source, publication date, article type, a one-sentence summary, the author's stance, the article's purpose, a list of information points, a list of entities mentioned, time sensitivity, and source quality. Tier two (Stage-2) is the interpretation tier: taking the fields extracted at tier one and reading them through a nine-dimension framework — patch and meta, tournament format, teams and players, regional landscape, club finance, rules and governance, risk profile, public narrative, and industry transmission.

The foundational principle is very simple: tier two is entirely dependent on tier one. It cannot recover information that tier one failed to extract.

That sounds obvious. But in practice, this obvious thing is violated every day, everywhere, in every industry — not just esports.

I remember June 2026, during the World Cup in Russia. I was following all 64 matches from Jakarta, writing for my personal blog. Germany lost 0-2 to South Korea, and their total xG in that match reached only 1.2 — the lowest in the national team's World Cup history. I wrote an analysis titled "The Collapse of a System," in which I pointed out that the PPDA metric of the gegenpressing mechanism the Germans pursued had fallen 23 percent compared with the 2026 World Cup. The piece was shared 15,000 times. An ESPN journalist reached out and invited me to contribute to a data column.

But what I took away from that experience was not "wow, I'm good." What I took away was this: the 2026 World Cup did not break my model; it expanded the definition of data. I realized that a good model is not a model that is always right, but a model that knows how to say "I don't know" when it truly does not know.

And this is where my esports story meets my football story again.

In Indonesia, esports is a distinctive market. Mobile Legends, PUBG Mobile, Free Fire, Valorant, Dota 2 — every title has its own ecosystem, its own tournament system, its own monetization model, its own governance structure. An analysis of one game cannot be carried over to another. A conclusion about the meta of one mobile title cannot be used to talk about VALORANT. This is something everyone in this profession must engrave on their memory: esports analysis must anchor to a specific title first, because tournament structures, data metrics, business logic, and governance structures diverge completely across titles.

Skip the title-anchoring step, and everything after it collapses.

And that is exactly what happened with the thirteen-page file. No title was identified. No teams. No players. No tournaments. No information points. No sources.

Yet the frame still stood there, nine dimensions, each with three layers of questions, demanding that I write conclusions.

---

The frame and the trap

I want to show readers what that nine-dimension frame looks like, because understanding it is understanding the whole disease.

Dimension one: Patch and meta. Here the questions are: what did the new update change? Which champions are up, which are down? How are win rates and pick-ban rates shifting? Which teams benefit, which suffer? This is a dimension where analysis is nearly impossible without numbers. Without numbers, every answer is a guess.

Dimension two: Tournament system and format. Which tournament, which tier, run by whom, what bracket, how many matches, what qualification path, what schedule density. Format determines upset probability, determines the length of the preparation window, determines the fairness of the qualification path.

Dimension three: Teams and players. Paper strength, positional fit, chemistry, bench depth, individual form, age curves, injury history, contract years. All of it requires at least one specific name.

Dimension four: Regional landscape. Which regions are strong, which are weak, import flows, academy quality, ecosystem health. And crucially: regional strength is title-specific — a region's standing in one game does not carry over to a different title.

Dimension five: Finance and business. Sponsorship revenue, league distributions, salary expenses, capital injections, contract structure, transfer fees. This is the dimension where silence is most dangerous, because in this industry, financial distress signals — unpaid wages in particular — appear at high frequency and must be actively checked.

Dimension six: Rules and governance. Competitive integrity, transfer and registration rules, contract compliance, minor protection, publisher governance controversies. This is the highest-severity dimension. If an article contains content about match-fixing, account boosting, contract disputes, or rule changes — and tier one drops it — that is a serious extraction failure.

Dimension seven: Risk profile. Competitive risk, financial risk, personnel risk, rules risk, public-opinion risk, and systemic risk. Each risk needs a level, a probability, an impact, and a mitigation.

Dimension eight: Public narrative and expectation. What story is being told? Does it have a fundamental basis? Is the sample size sufficient? What is the gap between market expectation and objective assessment? This is the dimension that analyzes things like "new king crowned," "dynasty succession," "all-domestic roster," "revenge arc," "veteran's last dance."

Dimension nine: Industry transmission. From upstream (publishers, event licensing) through midstream (clubs, tournaments, streaming platforms) to downstream (sponsorship, derivatives, mainstreaming).

Nine dimensions. Each one needs data. Without data, each one is a hole.

And the trap lies here: when an analytical frame is designed to demand conclusions in every dimension, it creates structural pressure that forces the analyst to fabricate conclusions. Fabricated patch notes. Fabricated roster moves. Fabricated financial signals. Fabricated narratives.

This is the kind of risk I call "analytical contamination through fabrication pressure." It is dangerous because it does not come from laziness. It comes from professionalism misplaced.

---

Patch and meta: when there are no numbers, silence is the only answer

I once worked with an esports team in Jakarta back in 2026. They played a MOBA title, and their analysis team had a habit I hated: after every patch, they would write a three-page "meta report" that always concluded that "this patch reinforces the map-control playstyle." No win-rate data. No pick-ban data. No comparison with the previous patch. Just feelings expressed in analytical language.

I asked them: "Where are the numbers?" They said: "The sample is not large enough yet." I asked further: "Then why does the report still have a conclusion?" They went silent.

That silence was the real answer. They did not write the report because they had a conclusion. They wrote the report because the frame demanded that a report must have a conclusion.

In patch analysis, there are three minimum data requirements you cannot do without if you want to say anything: win-rate data by champion or by weapon, pick-ban data, and a baseline compared with the previous patch. Without these three, every judgment about "the direction of the meta" is just storytelling. Storytelling is not wrong. Storytelling has value. But storytelling must be called by its proper name, not sold as analysis.

In the patch dimension, there is a subtler trap: the divergence between tournament servers and live servers. A patch may have gone live on the public server while a tournament is still locked to an older version. If you analyze the meta based on the live server while the tournament runs a different version, every conclusion you reach is systematically worthless — not because you are wrong, but because you are analyzing a different world.

For my thirteen-page file, this dimension contained nothing. No title, no version, no champion changes, no numbers. So the only honest answer is: insufficient information to assess.

Not "low risk." Not "stable." But indeterminate. The difference between "indeterminate" and "low risk" is the difference between a doctor saying "I don't have the test results yet" and a doctor saying "you're healthy." The second doctor is lying if he never tested you.

---

Tournament format: the shape of the frame decides who can win

I hold a fairly firm professional belief: format is one of the most underrated variables in sports analysis.

Take a football example I have watched for years. A two-legged tie is completely different from a single-elimination knockout match. In a two-legged format, the favorite has more time to correct mistakes. In a single match, upset probability is markedly higher, because the variance of a single match is far larger than the variance of two matches combined. Same team, same squad — change the format and you change the championship odds.

Esports is the same, even more visibly. A double round-robin, an upper-and-lower bracket, a single-elimination knockout, a points-based format — each creates a different probability distribution. In MOBA and tactical-shooter titles, the number of games in a series (best-of-three, best-of-five, best-of-seven) directly affects which team has the edge. A team with good roster depth but a slow start will prefer long series. A team with a few exceptionally strong strategies that are easy to counter will prefer short series.

Then there is schedule density. A tournament cramming three matches into four days is completely different from one spreading them evenly over two weeks. Dense scheduling creates a new scarce resource: recovery time and opponent-preparation time. Teams with strong analytical staffs benefit from gaps between matches. Teams that depend on reflexes and instinct are penalized.

The qualification path is also a variable of fairness. Some qualification paths are designed to protect big teams. Some are genuinely open. This difference is not only a matter of sporting ethics, but a matter of structure: it decides who actually has a chance of reaching the finals.

In my file, there was no tournament name. No tier. No organizer. No bracket. Nothing to say.

And here is what I want readers to remember: an analysis with no tournament name cannot establish the tier and nature of the event. It also cannot assess upset probability, the adequacy of the preparation window, or the fairness of the qualification path. You cannot analyze a frame when you do not know its shape.

---

Teams and players: where does the form curve sit inside an empty table

This is the dimension where I have the most professional memories.

March 2026. I was twenty-four, working as an assistant analyst at a football club in Jakarta. In a domestic league match against a strong opponent, I discovered a young midfielder named Septian David Maulana who had run only 8.2 kilometers — below the team average — but had made 11 passes into the opponent's final third, the highest on the team. He ran less, but he ran to the right places.

I presented a forty-page report proposing to move him from the wing into a number-ten role. The head coach dismissed it at first. After three trial matches, Maulana scored two goals and assisted three. The team won four straight. I learned something: data never lies, only the way it is presented decides how persuasive it is.

But I learned something in the opposite direction too, and that opposite thing matters more. If I had not had the distance-run data and the final-third passing data that day, I would have had nothing. I would have had only the feeling that "this kid plays well." And the feeling that "this kid plays well" — even when correct — cannot change a head coach's decision.

A player's value is not on the contract; it is in every off-ball movement. But to see those off-ball movements, you need positional tracking data, or at least event data. If all you have is a name and nothing else, you cannot analyze anything about that name.

That is the problem with the thirteen-page file. No team was named. No player. No role. No personnel decision — no signing, no release, no loan, no academy promotion, no retirement, no comeback. No contract status. No recent performance data. No coaching change.

In that situation, you cannot classify the roster phase — stable, adjusting, or rebuilding. You cannot begin form-curve analysis, age-curve analysis, injury-history analysis, or contract-year analysis. You cannot check whether the team depends on a single carry. You cannot check the divergence between a player's commercial value and his competitive value.

I remember a time in another domestic league when a club signed a player because he had a huge social media following. Jersey revenue went up. But his professional contribution metrics were not among the team's best. The board was happy with the commercial numbers. The coaching staff was troubled by the competitive numbers. That is a real paradox, and it is only visible when you have both types of data and know how to separate them.

With my file, both types of data were empty.

---

Regional landscape: why you cannot talk about a "strong region" without talking about a title

This is the most easily misunderstood dimension.

People often say: "Southeast Asia is strong in esports." That sentence is meaningless without specifying a title. Southeast Asia may be very strong in a mobile MOBA title, fairly weak in a tactical shooter, and middling in a traditional strategy title. Regional strength is title-specific. A region's standing in one game does not carry over to another title.

So when I read a report that talks about "regional strength" without anchoring to a specific title, I immediately know the writer is either lazy or hiding a data deficit.

Four pillars to examine when assessing the regional landscape:

First, international results. Not once, but as a trend across multiple events. A region winning once may be luck. Winning three out of four events is a different signal of quality.

Second, talent pool. The number of players at top-tier level. A region with twenty top-tier players is different from a region with two top-tier players and eighteen below the threshold.

Third, academy output. A region that produces its own players is more sustainable than a region that only buys.

Fourth, ecosystem health. Number of clubs, franchise-slot trading, viewership trends, publisher resource allocation.

Then there is talent movement. In many titles, larger regions import players from smaller regions. This flow is governed by import-slot policy. Once the policy changes, the flow reverses, and the regional landscape changes with it — within a single season.

In my file, no region was named. No tournament. No transfer flow. Nothing.

And here is the key point: you cannot establish regional tiering without a title. That is not excessive caution. It is the basic logic of the craft.

---

Finance and business: silence is not innocence

This is the dimension where I want to issue my strongest warning.

In esports, there is an uncomfortable truth: unpaid wages are a high-frequency distress signal. Young clubs, lower-tier teams, emerging organizations — all of them can fall into delayed wages, unpaid wages, dissolution, or franchise-slot sales. These events are rarely publicly announced. They tend to leak through social media, through players' accounts, through status posts deleted after a few hours.

So when an analytical report contains no financial signals at all, you are not permitted to read that silence as "no risk." The absence of a signal is not evidence of absence. This is a null field, not a clean bill of health.

An esports club's financial structure typically has four main sources:

Sponsorship revenue. This is usually the largest source, and also the most risk-concentrated. Losing a title sponsor can collapse an entire season's budget.

League or publisher distributions. This source depends on position within the system — top teams get more, lower-tier teams get little or nothing.

Salary expenses. In this industry, player and coach salaries usually make up a large share of total costs. The salary-to-revenue ratio is an important health indicator.

Capital injected by investors or owners. Many clubs survive on this flow, and when the flow stops, everything stops.

When assessing a transaction — a signing, a renewal, a transfer — you need at least one number. What is the transfer fee. What is the salary. How long is the contract. Is there a release clause. Without these numbers, you cannot judge whether the price is high or low relative to true value.

And this is what I have warned colleagues about many times: if a source article contained signals of unpaid wages, match-fixing, injuries, or rule changes — and the extraction tier dropped those signals — that is a serious extraction failure that must be re-run. Because these are precisely the highest-severity content categories, and they must never be allowed to disappear silently.

With my file, no financial event was identified. Revenue-concentration analysis, salary-to-revenue analysis, franchise-slot amortization, and overpricing arms-race analysis — none of them applied. Not because they are meaningless, but because they had no subject.

---

Rules and governance: the highest-severity dimension

If you ask me which of the nine dimensions I never allow myself to skip, the answer is this one.

The reason is simple. In esports, governance authority belongs to the game publisher. The publisher decides the rules of play, the schedule, the transfer rules, the penalties. So you cannot identify the applicable rules system if you do not know the title. That is an unbreakable dependency chain.

The checklist in this dimension usually includes:

Competitive integrity. Are there signs of match-fixing, result manipulation, intentional losing? This is the most severe content category, and once it appears, it must be escalated to top priority.

Transfer and registration rules. Does a transfer violate a transfer window? Does it violate import-slot rules? Does it violate player registration conditions?

Contract compliance. Is there a dispute between player and club? Is there a penalty clause that has been breached? Is there a "contract prison" — a long contract and a large buyout effectively blocking a transfer?

Minor protection. This is a morally and legally sensitive dimension, and it is often omitted in commercial reports.

Publisher governance controversies. Has the publisher changed a rule in a controversial way? Has it penalized a team for a controversial reason?

When you have an alleged violation, you can build three scenarios: worst case, middle case, optimistic case. These three scenarios help readers understand the risk range.

But in my file, there was no alleged violation at all. No integrity issue. No transfer-compliance issue. No minor-protection issue. No governance controversy.

And I must emphasize: that does not mean clean. It means there was nothing to assess.

---

Risk profile: the real risk is not in the article, it is in the process

This is the dimension where I want to linger longest, because it holds the biggest lesson.

In a standard risk profile, you have six groups: competitive risk, financial risk, personnel risk, rules risk, public-opinion risk, and systemic risk. With my file, all six content groups were indeterminate — not low, but indeterminate, because there was no subject.

But there is a genuine systemic risk, high-probability, high-impact: the risk that a completely empty tier-two report gets consumed downstream as though it were real analysis.

I call it the risk of a frame that knows how to dress itself up.

Imagine this. A process outputs a thirteen-page document with clear section headers, tidy tables, professional language. The downstream reader — a coach, a sporting director, an investor — skims it, sees the full structure, and assumes the source article was analyzed. But the source article was never read. What they are reading is a frame filled with descriptions of missing data, which superficially resembles an analysis.

In sports, we have seen the same thing. Scouting reports dozens of pages long, full of charts — but if you peel back the charts, you find the entire conclusion rests on fifteen minutes of video of a twenty-year-old player. Match-analysis reports six pages long — but if you separate out the numbers, you find they come from a sample of three matches, and the third match has not yet finished.

The danger is not that people use small samples. The danger is that people do not say the sample is small.

In my profession, there is a sentence I always remind myself of: my model is only bad when I am too cowardly to ask it the hardest question. The hardest question is not "is this team strong." The hardest question is "do I have enough data to answer that question, and if not, do I dare to say no."

With my file, the correct answer is: insufficient data, cannot assess, and the extraction tier must be re-run before any decision use.

---

Public narrative: when a small sample wears a king's robe

I once followed a phenomenon I think everyone in this profession encounters.

A team wins three matches in a row. The media starts calling them a "sensation." Fans start talking about the title. Sponsors start showing interest. In the fourth match, they lose. In the fifth, they draw. By the sixth, the story has vanished from the papers, and so have the sponsors.

I call that a story with no fundamental basis. Three matches is far too small a sample to say anything about a team's quality. But three matches is a perfect sample for generating a media narrative.

Public narratives in sports tend to follow a few familiar templates: new king crowned, dynasty succession, all-domestic roster, revenge arc, veteran's last dance, career comeback. Each template has a different life cycle, and each template has a different way of testing its own sustainability.

When assessing expectations, you need three columns: market expectation, your objective assessment, and the gap between the two. If the gap is large, either you are right and the market is wrong, or you are wrong and the market is right. There is no third column to hide in.

Then there are sentiment indicators. The ratio between social media heat and fundamentals. If heat is high and fundamentals are low, you are looking at a bubble. A bubble can keep inflating for a while. But eventually it bursts, and the person who bursts with it is the one who believed in it last.

Where a player is retiring or returning, you need to judge motive — form decline, injury, business transition, or contract dispute — and assess the feasibility of a return. Without those, you are just writing a moving story, not an analysis.

And in my file, there was no subject at all. No player, no team, no narrative tag, no sentiment data, no expectation signal.

---

Industry transmission: three tiers of one flow

The final dimension is the one I like most intellectually, because it forces you to look at a whole system rather than a point.

The transmission flow runs from upstream to downstream.

Upstream is the game publisher, event licensing, patch cycles. A publisher changes one line in a tournament organization clause, and an entire ecosystem shifts.

Midstream is clubs, tournaments, streaming platforms. This is where the flow is converted into salaries, contracts, and content.

Downstream is sponsorship, derivatives, mainstreaming. This is where esports meets the mainstream world — through multi-sport continental events, through multi-title international events, through capital from regions with strategic ambitions.

At each tier, you can look at a different industry:

Game publishers. What is their strategy? Are they expanding or contracting? Which regions are they investing in?

Streaming ecosystem. How are broadcast rights being priced? Which platforms are paying the highest?

Sponsorship and marketing. Where is the sponsor structure shifting from and to?

Offline and derivatives. City naming rights, team naming rights, merchandise, offline events.

Mainstreaming. How much closer is esports moving to mainstream sporting institutions?

Gray zones. The betting market and other gray areas. Here I must be explicit: analyzing information objectively is one thing; giving betting advice is a completely different thing. We do not do the second.

And of course, with my file, all three tiers had no subject. There was no transmission path to trace.

---

The contrarian angle: the frame is not the solution, the frame is part of the problem

This is where I want to go against the reader's expectation.

When people talk about analytical quality, they usually think of three things: better data, better models, better experts. They believe the problem lies in the input.

I think the problem more often lies elsewhere: in the frame.

A well-designed analytical frame protects the writer from himself. It has blanks that allow writing "insufficient information." It has minimum thresholds — for example, at least five discrete, source-attributed information points, at least one identified title, at least one specific name — and if those thresholds are not met, the process halts instead of continuing.

But when a frame is designed to demand conclusions in every dimension, it turns honesty into an impossible behavior. The writer is placed in a bind: either fabricate, or admit that he cannot do the job. And in a normal work environment, people usually choose fabrication, because fabrication looks like competence while admission looks like inadequacy.

This is a structural paradox. The frame built to guarantee quality becomes an incentive to produce false information.

In football, I have seen the same thing at a different scale. Clubs build elaborate metric systems, hire good analysts, buy expensive data packages. But when it comes time to make a transfer decision, they still sign the player the coach likes, and the metrics are used to rationalize a decision already made. The data is there, the frame is there, but the role of both has been reversed.

So my contrarian question is: if you want to improve analytical quality, are you willing to design a frame that allows it to fail?

A frame that allows it to fail is a frame that allows it to be honest. And a frame that allows it to be honest is the only frame worth building.

I remember a coach I once worked with. After every loss, he did not look for someone to blame. He opened the tape, reviewed it, and took notes. A good head coach treats a defeat as an update, not a verdict. He told me something I have carried through my whole career: "If your data says I am wrong, I want to hear it early, not late."

An empty report is also data. It tells you your process is broken. The only question is whether you are willing to listen.

---

What actually needs to happen

If I have to close with a constructive view, here is what I think the sports analytics industry — football and esports alike — needs to do better.

A hard rule is needed: an empty file must never be passed from the extraction tier to the interpretation tier. If the extraction tier cannot identify the article title, source, and type — three fields any retrievable document can supply — that is a sign of a process failure, not of an article with no content. And a process failure must be fixed, not dressed up in prose.

A hard gate is needed for the title. If the title cannot be identified, no tier behind it is permitted to operate. This is a necessary condition, not a sufficient one, but it is a non-negotiable condition.

A clear checklist is needed for high-severity content categories: unpaid wages, match-fixing, injuries, rule changes, minor protection. These categories must never be allowed to vanish silently through processing steps. If they vanish, the process itself must raise an alarm.

A minimum threshold is needed for the number of discrete, source-attributed information points before the interpretation tier is allowed to run. An analysis based on two information points and an analysis based on twenty information points are not the same product, and they should not look the same when printed.

And finally, a cultural change is needed. In this profession, we reward confidence. We reward the person willing to make a bold prediction. We rarely reward the person who says "I don't know." But the truth is, in data analysis, "I don't know" is one of the most valuable answers you can give — because it is the only answer that is correct when it is correct.

I remember March 2026, when global tournaments were suspended because of the pandemic. I was twenty-seven then, head of the data department at another club in Indonesia. I built a report on the effects of playing without crowds on performance, proposing to increase high-intensity running distance by 12 percent to offset the lost home advantage. When the league returned in October 2026, my team went eight matches unbeaten — the best run in the club's history. The coaching staff called me "the mad professor."

But what I remember most is not those eight unbeaten matches. What I remember most is the period before, when I had no data — because there were no matches to pull data from. In that period, I learned that data is not something to display. It is a tool for surviving a crisis. And in a crisis, the first tool you need is honesty about what you have and what you lack.

---

A forward-looking closing

I do not know what the source article behind that thirteen-page file was said to be analyzing. It might have been a major transfer. It might have been a governance scandal. It might have been a tournament reform. It might just as easily have been an unremarkable piece about an unremarkable match.

I will never know, until the process is re-run and the fields are properly populated.

But I know one thing: data never lies — only the way we listen is wrong. And when there is no data to listen to, the only way not to lie is to say that there is nothing to hear.

In this annual season — whether a football season or an esports season — readers will continue to receive hundreds of analyses a week. Some will be built on real data. Some will be built on feelings given analytical names. And some will be built on empty frames filled with confidence.

The question I leave readers with is not "which analysis do you believe." The question is: when you read an analysis, do you ever ask how many source-attributed information points it was built from? And if the answer is no, then who is analyzing whom?