Esports and the Data Problem: Why One Framework Cannot Analyze Every Game Title

7 p.m. Riyadh time, mid-July. In the analysis room of a multi-title tournament, five screens show five matches running side by side: a League of Legends game, a Dota 2 game, a Counter-Strike 2 game, a Valorant game and a PUBG Mobile game. Five matches, five metric sets, not a single column of shared data. On the MOBA side, analysts measure objective control rate and lane-push tempo. On the FPS side, they measure opening-duel win rate and kills per round. Same event, same match day, but two entirely separate frames of reference. When data speaks, the whole stadium has to fall silent — but here, each stadium speaks its own language.

That is a miniature portrait of a much larger problem: the esports analysis industry is growing faster than its own measurement infrastructure. Watching hundreds of matches from regional to international level over six years, I have repeatedly seen analysis teams, media outlets and even professional data companies try to build a single "universal" framework for the whole industry. And repeatedly fail. The cause is not a lack of data — this industry has too much data. The cause is that each title's data runs on its own logic, and those logics cannot be blended.

Why esports has no shared metric set

Esports is an umbrella covering at least five groups of titles with fundamentally different competitive structures: MOBA (League of Legends, Dota 2, Honor of Kings), tactical FPS (Counter-Strike 2, Valorant), battle royale and tactical arena (PUBG Mobile, Free Fire), simulation sports, and fighting games. Each group has its own unit of time, unit of victory, and unit of performance measurement.

In MOBA, the value of a play is measured by accumulated advantage over time: gold, experience, objective control. In tactical FPS, value is measured by survival rate per round and by bomb-plant or site-control situations. In battle royale, value depends on final placement and kill count, two variables that frequently contradict each other. A strong player in one group cannot be judged by the yardstick of another.

Even within the same group, titles do not share metrics. League of Legends and Dota 2 are both MOBAs, but their map structures, item systems and positional roles differ enough to disable any direct comparison. Faker — Lee Sang-hyeok — is regarded as the icon of League of Legends, but his record cannot be translated to Dota 2 or any other title. Likewise, a top-tier marksman in Counter-Strike 2 such as ZywOo or s1mple cannot be assessed with the same metrics in Valorant, even though both are tactical FPS.

I once tried to apply an expected-goal framework — one familiar from football — to tactical FPS matches. The result was a string of meaningless values. "Chance conversion" in FPS does not operate as it does in football, because each round is a closed game that can end without any "chance" being created in the attacking sense. The lesson: an analytical tool cannot be separated from the rules that produced it. I do not commentate on esports. I read it through charts — but each chart must be drawn on the right kind of paper for that title.

Individual metrics and the lone-hero trap

In any team title, judging an individual in isolation from the collective is a near-impossible task. A player may own a beautiful individual stat line because teammates create conditions for them, or a poor one because they are carrying a sacrificial role. In MOBA, the jungler often has a lower damage stat but is the one who sets the tempo of the match. In tactical FPS, the in-game leader — IGL — may have a low kill rate but is the one who draws the entire tactical structure. Read only the scoreboard, and you will misjudge both.

This is why I always place individual metrics in the context of role. A metric separated from role context is just a dot on a chart, with no analytical weight. And in an industry where teams increasingly specialise their roles, reading that context correctly becomes the core skill of an analyst.

Patches and the life cycle of the meta — where old data becomes poison

One feature makes esports fundamentally different from traditional sports: the pace of change. In football, the rules have been nearly static for decades; a data model built in 2026 still has reference value in 2026. In esports, a publisher can push a balance update every two weeks, changing champion strength, items or maps. Each time it does, a portion of historical data loses value.

This is where esports analysts fall into the trap most easily. Old data is verified, safe, easy to cite. But in a constantly shifting meta environment, that safety is an illusion. A composition that dominated last season can collapse after a single patch. A player with impressive metrics can lose their rhythm once their role is redefined.

The rule I set for myself: every esports analysis must draw at least forty percent of its data from the current period. This is a harsh discipline, but a necessary one, because audiences are more fascinated by historical values than by a reality in motion.

Tournaments and formats — the forgotten variable

Competitive format is an undervalued variable in most esports analysis. A best-of-three series differs fundamentally from a best-of-five, not only in length but in the probability of an upset. A single-game format sharply increases variance, giving weaker teams a greater chance. Conversely, double-elimination or Swiss formats reward consistency.

When analysing a team, I always ask: which format are they playing, and how much of their result was shaped by bracket luck. A team that reached the playoffs through an easy bracket should not be rated level with one that survived a bracket of death. But in fast news briefs, this difference is often erased, and every win is treated the same.

Tournament structure also determines the weight of each result. Multi-title events gathering more than twenty disciplines, most notably the Esports World Cup, pose an unprecedented challenge: one organiser, one broadcast frame, but dozens of parallel metric systems. The transformation of events into commercial products with fixed franchising slots and no promotion or relegation has changed competitive incentives in ways pure data struggles to reflect fully. When a slot is bought rather than earned, competitive pressure at the bottom of the table falls, while commercial pressure at the top rises.

Schedule density and the limits of stamina

Match density is an increasingly important variable but is rarely built into prediction models. A team that must travel between continents and play for weeks on end will perform differently from one that is fully rested. At the professional level, the gap between top teams is often very small, and stamina can reverse a result. But there is no standard data column for "exhaustion level," so it is often omitted from analytical tables.

Regional strength — a non-uniform map

The map of regional strength in esports is also a problem with no common answer. A region can be number one in one title but a complete outsider in another. South Korea has dominated League of Legends for years, but in Counter-Strike 2, Europe is the centre. Southeast Asia is strong in mobile battle royale titles, while North America holds advantages in some FPS titles.

This makes any claim like "region X is an esports power" meaningless unless tied to a specific title. Cross-regional transfer flows make the picture more complex still. A player moving from one region to another changes not only the jersey colour but the entire training environment, match calendar and way of reading the game. Importing players can fill a talent gap immediately, but it also masks holes in the youth development system — a problem that only surfaces after several seasons.

Team economics — where data reflects reality

At the economic layer, esports operates like a young market with wide swings. A team's revenue may come from sponsorship, publisher revenue sharing, in-game item sales or prize money. Over-reliance on a single source is a red flag. When a team loses its main sponsor, or when a publisher cuts revenue sharing, the consequences can arrive faster than anyone predicted.

The industry's most dangerous signal — unpaid wages — almost always appears before dissolutions or sales. It is the most weighty information but also the hardest to verify, because the parties involved have incentives to hide it. In esports financial analysis, I learned to separate investment value from liquidation value — two figures that are often confused but reflect two entirely different stories about an organisation's health. Transfers are a market, and markets have no emotions — only liquidation value and investment value.

Governance and compliance — the grey zone of the rulebook

Each title has a publisher with its own rulebook, and these rules not only differ but sometimes conflict. A behaviour banned in one title may be accepted in another. Transfer issues, player registration, minor protection and competitive integrity all sit under a fragmented governance system. This creates a paradox: the more professional the industry becomes, the greater the demand for a unified governance framework, yet power rests with individual publishers whose interests do not always align. When disputes arise, there is no shared court to arbitrate — only the terms of service of each platform.

A counter-intuitive angle

Here is a paradox I want to put on the table: the fragmentation of esports data, which many treat as a weakness, may be a strategic strength. When each title owns its own metric system, it forces analysts to genuinely understand the game rather than apply a ready-made formula.

The traditional sports data industry once fell into the opposite trap: overconfidence in universal probability models to the point of ignoring specificity. The model predicting France to win Euro 2026 based on expected goals was wrong, while a team with a lower metric lifted the trophy thanks to transcendent individual talent and football's inherent uncertainty. Esports, through its fragmentation, is partly immune to that kind of arrogance. Here, no one can pretend to understand everything.

But correlation does not mean causation. A team having better metrics does not mean those metrics create victory. It may be the winning style of play that generates the pretty metrics, not the reverse. A careful analyst must always ask: is this the cause, or just the consequence seen in a mirror?

The limits of data

I have to admit one thing: data never tells the whole story. It cannot measure psychological pressure in front of a packed stand, nor the emptiness of a stadium without spectators — something the 2026 pandemic laid bare for both football and esports. The empty stadiums of 2026 stripped modern sport bare: no crowd, no roar, only data speaking in place of everything. When the roar disappeared, home win rates fell and playing behaviour changed. Those variables sit in no data table, but they exist, and ignoring them is an analytical error.

Every analysis I write must include a section on the limits of data. Not as self-defence, but to remind readers that data is a tool, not a truth. In the transfer market, every deal can be priced with cold figures, but the final decision still belongs to people — and people cannot be fully measured.

Takeaway

The question for the next cycle is not how to standardise esports data across the whole industry, but how each title ecosystem can build a metric set deep enough while retaining the ability to move between those ecosystems. Esports will not have a common language in the near future. But it can have a multilingual dictionary — provided analysts are willing to learn each language one at a time, instead of hoping for a universal translation.