When the Base Data Layer Is Empty: Lessons From an Esports Analysis That Could Not Be Completed

At the fourteenth minute of game two, the two teams stood nearly two thousand gold apart and neither dared open a fight. On the broadcast's analysis board, a number appeared: a 78% pick-ban rate. I muted the commentary, opened three tabs, checked four public data sources, and found no table anywhere that confirmed it. Nineteen minutes later, one team lost. The cell stayed on screen. Nobody verified it. Nobody took it down.

I wrote that moment into my notebook, beside a small line: source — unidentified.

That seemingly trivial detail is why I sat for eleven hours one January night in front of an unfinished spreadsheet. I was reconstructing an entire regular season of Southeast Asian esports: from patch versions, formats and rosters to transfers, club finances and disputes off the server. My spreadsheet has nine layers. The first layer was empty.

When the base data layer is empty, every layer above it stops being analysis — it becomes an assumption delivered in the voice of fact.

A season needs data, and a region does not supply enough

My work is unremarkable. In seven years living in Brisbane, I have made a living turning esports matches into tables: who banned what, who picked what, where the gold went in the first ten minutes, which team changed its split-push timing after taking a Herald. For League of Legends, the raw material is reasonably complete. Even a region like Southeast Asia offers win rates by role, gold differential at fifteen minutes, and win rate after securing the first major objective.

The problem lies elsewhere.

Southeast Asian esports does not live on League of Legends. It lives on mobile titles. Arena of Valor, PUBG Mobile, Free Fire, Wild Rift — those are the names that hold audiences in Vietnam, Thailand, Indonesia and the Philippines. And those are precisely the titles with the thinnest data infrastructure. For a regional mobile match, what I usually receive is a final scoreboard: kills, creep score, damage dealt. No timestamps, no heat maps, no second-by-second fight logs.

That mismatch became clearer after two milestones. The 31st SEA Games in Hanoi in 2026 put esports into the official competition programme, and the 2026 Asian Games in Hangzhou turned esports into a genuine medal event. Mainstream newsrooms began asking questions that only hardcore fans used to ask: where is this team strong, why did they win, why did they lose. Demand surged. Supply barely moved.

I call it the empty-cell syndrome. When a newsroom needs a number and no number exists, the natural reflex is to take the nearest approximation and call it by an exact name. A visual count becomes a percentage. A good match becomes a regional trend. One week of play becomes a tactical identity.

The nine layers of an analytical table

When reconstructing a season, I always work through nine layers, from root to canopy. The order is not ritual; it is the condition under which any conclusion above can still stand.

Layer one — patch and meta. This is the root layer. A small numerical change can reverse an entire draft approach. With League of Legends, I can track each patch cycle and cross-check the date a tournament switched to the competitive server build. With a regional mobile title, I often have nothing but an update announcement in a local language and a few feel-comparison videos. Without layer one, every conclusion above is a decorated guess.

Layer two — tournament format. Format decides upset probability. A single-game series and a best-of-three are different worlds in terms of how stable a strong team looks. I once rebuilt an entire regional qualifier just to answer one question: how much of the champion's run came from format, and how much from genuine strength. The answer was uncomfortable — under short formats, upset rates rise sharply and every power ranking loses value.

Layer three — rosters and players. This is the richest layer and also the easiest to distort. In League of Legends I can compare gold differential, creep score and fight participation for individual players week by week. In mobile titles I usually get only the final kill count — a metric anyone who has watched a match knows is close to meaningless without context. A player forced onto a losing side lane can end with a prettier scoreline than a top laner who absorbed pressure all game.

Layer four — regional landscape. Southeast Asia is not one bloc. Vietnam, Thailand, Indonesia and the Philippines each have their own ecosystem, their own competitive rhythm, and their own flagship titles. Yet regional comparisons often take international results and infer the quality of the whole base. That is a logical leap requiring supporting data: youth pipeline size, number of domestic events, stability of the competitive calendar. I tried that inference for three seasons, and each time I had to go back and revise it for lack of sample.

Layer five — club finance. Here data is not only sparse but sensitive. The 2026 Esports World Cup in Riyadh, Saudi Arabia, ran from July to August with a total prize pool above 60 million US dollars. That figure reshaped the income baseline of organisations across the board. But when I needed to know how many Southeast Asian players a given team sent there, what the trip cost, and how the winnings were redistributed — I had no source reliable enough. Half the picture sat out of reach.

Layer six — rules and governance. This is the layer I approach most cautiously. Contracts, transfers, player registration, and issues involving underage competitors all live here. Without primary documents, I do not conclude. Experience has taught me that an article built on a rumour about contract violations can cause real harm to a real person, while the author pays nothing.

Layer seven — risk profile. I build a risk table for every team: competitive risk, financial risk, personnel risk, public-opinion risk. My rule is simple: an empty cell is written as “insufficient information”, never as “no risk present”. Those two statements differ in substance, and confusing them has produced no small number of false reports over the years.

Layer eight — public narrative. This is the layer Southeast Asian esports generates most prolifically. New king stories, dynasties ending, farewells, heroes returning. Most of them have short lifespans because they lack a data foundation. A team wins three straight matches and is immediately labelled a title contender; three weeks later the same team loses four and vanishes from every prediction table.

Layer nine — industry transmission. From publisher, through clubs, tournaments and streaming platforms, to sponsors and derivative markets. Every shift in layer one flows down here, just more slowly, over months. I once tracked a fairly small mechanic change in a mobile title, and by the following season the transfer lists in two countries looked entirely different. Nobody named the cause, but the transmission line could be drawn.

The trap sits in layer one, not in the conclusion

During those eleven hours, I uncovered something fairly unexpected about how this industry gets its analysis wrong.

Errors rarely sit in the conclusion. People usually conclude quite sensibly. Errors sit in layer one — where patches, server builds and mechanic changes are skipped because they are too hard to verify. An analysis claiming Team A presses better than Team B may be correct, but if Team A played on an older patch against weaker opponents, that correct conclusion rests on a false floor.

When the base data layer does not exist, the right question is not “which team is stronger”, but “do I have the standing to conclude at all”.

In sports data analytics, people talk about margin of error, sample bias, minimum sample size. But in Southeast Asian esports, the first problem is the existence of data. There are tournaments where I cannot find an official schedule in a machine-readable format. There are matches where the final scoreboard on the official site differs from the one on broadcast. There are events where the competitive patch version is never published, rendering any week-over-week comparison meaningless.

I once spent nearly a month establishing that a regional league switched to a new patch two weeks later than usual. Two weeks sounds small. But if you use data from those two weeks to compare against the rest of the season, you are comparing two different things and calling them by the same name.

Every number carries a story, and my job is not to ruin it.

Counterintuitive: more data is not the answer

At sports conferences I hear the same line: we need more data. I do not believe that is the right answer in this region, at least not for several years.

More data without source discipline will only produce more empty cells filled with guesses — guesses that merely look more professional because software sits behind them. The scarce skill in Southeast Asia is not modelling. The scarce skill is the discipline to declare a null — to write that there is insufficient information to assess, to take a chart down when its provenance does not hold.

At 39, I have learned that data also hurts when it is distorted.

But I must tell the other half of the story, because telling only the first half would be lying to myself.

Some things tables cannot measure, and in mobile esports they make up most of the match. How a player moves inside a teamfight, when a player changes direction, the calm of a captain while his team is behind — no metric captures those. I once spent two nights breaking down a single acceleration by Kylian Mbappe in France against Argentina at the 2026 World Cup round of sixteen, frame by frame. None of my pressing and expected-goal numbers explained the raw beauty of that run. Data measures what happens, not what makes people love the game. The same holds for esports.

So the second counterintuitive point is harder to swallow: sometimes a commentator sees better than a spreadsheet, because the human eye catches correlations a model has not been taught to look for. But that sight must be spoken as an observation, never as a statistic. That is the line I have held throughout my career: if you speak with your eyes, call it an observation; if you speak with a table, you must have the table.

Signals for the next season cycle

When the table speaks, the stadium must learn to stay silent. But before the table can speak, the practitioner must learn to stay silent in front of empty cells.

Next season cycle, I will not be tracking which team wins first. I will be tracking which organisation voluntarily publishes its match data — full draft lists, objective timestamps, competitive patch version. That is the earliest signal of a maturing esports ecosystem, and it does not depend on whether that team wins or loses this week.

A tournament can buy a star. An esports ecosystem can only buy transparency by volunteering to publish. And when an esports ecosystem begins to publish, people like me, sitting behind a computer, finally have work worth doing.