Southeast Asian Esports and the Data Void: Unpaid Wages, Empty Franchise Slots

Southeast Asian Esports and the Data Void: Unpaid Wages, Empty Franchise Slots

On the night of January 8, a League of Legends team in Southeast Asia posted a new jungler announcement on its homepage. I opened my transfer-tracking spreadsheet, a file that has been running continuously since June of last year, and typed the player's name into column A. Columns B through G were blank: base salary, contract length, release clause, buyout fee, effective date, performance bonuses. The only filled cell was column H, where the team's communications office had written two words: "undisclosed".

All I could fill in were the traces public matches leave behind: participation rate in early-game skirmishes within the first 10 minutes, gold differential at minute 15, neutral objective control count, map pressure index. This new jungler sits in the bottom 12 percent of the region for early-map pressure, with 0.41 cross-map rotations per minute. I highlighted that cell in red and wrote a note beside it: needs twenty matches to confirm.

A contract is announced through a press release. Its substance lies in the blank cells nobody bothers to fill. Esports is running the largest talent economy in the history of competitive gaming on an empty database.

The transfer window runs on press releases, not numbers

Football has Transfermarkt. Basketball has publicly disclosed payrolls under collective bargaining agreements. Southeast Asian esports has... press releases, a few tweets from agents, and interviews in which players say the new team "really fits my direction".

I am not objecting to discretion. I am objecting to an industry declaring itself a professional sport while the three most basic variables of any professional labour market — contract length, contract value, termination terms — sit beyond public verification. A league can publish its total prize pool down to the dollar while concealing the salaries of every single person who produces that league.

Over the past eighteen months, I logged 214 transfer and contract-extension announcements across four regional competitive systems: League of Legends, Mobile Legends: Bang Bang, VALORANT and Dota 2. Of those 214 announcements, 31 included a specific figure. The disclosure rate was 14.5 percent. In 69 cases, the only phrase used to describe value was "undisclosed". The rest used vague formulations such as "long-term contract" or "multi-year commitment".

That 14.5 percent figure is not an esports-specific problem. It is the problem of an entire analytics industry trying to sell audiences forecasting models built on data that does not exist.

Three substitute indices when contracts are sealed

Without payrolls, an analyst is forced to build substitute indices. I use three main groups, and I state upfront that they carry error.

The first group is roster structure: average age, age standard deviation, number of players under publicly known contracts, number of players with international experience. This is the easiest group to verify because it sits inside competition records.

The second group is performance profiles broken into 15-minute blocks. I split each match into six blocks and calculate gold differential, objective differential and fight participation for each. A team can win 2-1 while losing the gold differential in the 45-to-60-minute block. The scoreboard does not reflect that. The dataset does.

The third group is mid-season roster churn frequency. I count how many times a team changes players between competition weeks. In my data from 2026 onward, teams that change rosters three or more times in a season have a 2.7 times higher probability of dropping out of playoff contention than stable rosters. The correlation is clear, but I will return to it in the contrarian section.

Numbers do not lie, but they do sulk. What they sulk about most is being used to fill a gap where the underlying data never existed.

The patch-learning curve: the most mispriced variable

Across the four competitive systems I track, League of Legends and VALORANT have markedly different update cycles. League of Legends rebalances every two weeks, with one major mid-season update. VALORANT moves more slowly, but each change usually targets a specific weapon group or agent.

I built an index called adaptation lag: the number of matches required for a player's individual performance metrics to return to his own baseline after a major update. Data from 41 professional players I tracked over two years shows an average lag of 6.9 matches, with a standard deviation of 3.4. The fastest group needs two matches. The slowest needs seventeen.

The economic meaning is direct: a player with an adaptation lag of fifteen matches spends almost an entire group stage performing below par, while his team still pays his full contractual salary. If a major update lands at the start of a season, the team that owns him loses roughly a third of the season.

No transfer announcement mentions this index. No negotiation, as far as I know, writes it into contract terms.

From the Leicester City season I tracked through 2026-2026, I learned something transferable to esports: when you lose two cornerstone figures at once and do not replace them with equivalent structure, the metrics warn you nine to fourteen weeks before the league table does. I call these leading indicators. Leicester collapsed before the table even noticed. Esports is no exception.

Invisible payroll and a warning from a football season

This is the section I consider most important, and also the hardest to verify.

In traditional professional sport, wage delays surface in the press before they surface in the standings. Players strike, coaches answer questions, players' associations speak up. In Southeast Asian esports, wage delays typically only become visible after a player has already left the team and posted about it on social media — that is, after the season has ended.

I tried to build an indirect index: the rate of sudden mid-season roster departures. Of 22 cases I recorded of players leaving teams abruptly mid-season since 2026, 14 were later followed by public information about delayed or unpaid wages. That is 63.6 percent. This is a correlation, not causal proof, and I will stress that point again at the end.

But the correlation says something about how the industry operates. If 63.6 percent of mid-season departures relate to cash flow, then analysing a team purely through in-game data is analysing half the truth.

This is why I always ask three questions when assessing a team: how long does the head coach's contract run, how many main sponsors has the team announced, and when did the team last publish financial statements. In esports, the answer to the third question is usually never.

When a publisher targets a role

Publishers do not deliberately sabotage anyone's career. But every major update redistributes player market value, and that redistribution is not priced into the transfer window.

Take a verifiable example: during a period when the publisher reduced the strength of the bruiser champion class in the top lane in League of Legends, the relative value of players who specialised in that class fell while the value of tank-specialist players rose. In the transfer market, both groups are usually still priced on last season's results — that is, on stale data.

This is the kind of systemic error I chase: a rule-governed mismatch between market price and practical value, repeating every transfer window, never corrected because nobody measures it.

The test is not complicated. For each player, I compare performance metrics over the last ten matches with last-ten-match metrics plus notes on the current patch version. If the gap exceeds one standard deviation, I flag it. Of the 60 players I tracked in the current transfer window, 17 fell into this group. Only four showed signs of being renegotiated in value.

The in-game leader: the most underpriced asset

On any roster, the in-game leader is the hardest factor to quantify and usually the factor paid below his contribution.

I try to measure him with three indirect indices: the deviation between fight-calling decisions and fight win rate, the density of shot-calling signals during blocks where an advantage was reversed, and the team's win rate in the five minutes after losing a major objective. The third is my favourite, because it measures recovery capacity after a shock event.

In my data, teams whose in-game leader scores one standard deviation above average on the recovery index have a 1.9 times higher comeback rate after losing a major objective. Yet when I compare publicly disclosed salaries — which are extremely few — this group shows no corresponding pay gap.

I stress the limitation: my sample is small, and salary data barely exists. I do not trust emotion, I trust systems — but I always check the systems. This is a case where I have to say plainly that my system is missing a leg.

Southeast Asia: a talent export factory and the trap of buy-low, sell-high

Our region runs a simple model: discover young talent, give them two seasons of competition, sell them to major leagues, reinvest in the next player.

This model works for short-term cash flow. It has three structural weaknesses.

The first weakness is compressed sale prices. When buyers know you do not disclose contract structure, they negotiate from a better information position. The buying side always knows more than the selling side about an asset's true value in the international market.

The second weakness is the loss of organisational skill. When you sell an in-game leader or an analytics coach, you lose not just an individual but a decision-making system. That loss never appears in a transfer report.

The third weakness is that turnover is so fast that leading indicators never accumulate. I have observed teams that change more than half their roster every season for three consecutive seasons. No indicator can forecast reliably on data that is continuously scrambled.

My first-hand experience watching VCS matches and regional competitions shows a recurring pattern: teams with outstanding young players tend to win more in the early season, when opponents lack data on them, and decline in the later stage once enough data has been collected. This is a pure information asymmetry effect, and it disappears if you have a decent internal data collection system.

Competitive integrity in a non-disclosure environment

This is the section I write with the most caution, because it touches specific individuals and organisations.

Esports betting has grown faster than the governing bodies' capacity to monitor it. The concern is not whether betting exists. The concern is that football's open-data environment — where every pass, every run, every minute is recorded and published — is precisely what enables the detection of anomalous patterns. Esports has far richer in-game data, yet lacks the second data layer: contracts, wages, internal conditions, relationships between parties.

With only in-game data, you can detect an anomaly but not explain it. An underperforming player might be underpaid, in internal conflict, dealing with personal issues, or something else entirely. Without the second data layer, every conclusion stays at the level of speculation.

I make no accusations here. I make an infrastructure demand: any league that wants to protect its integrity against betting markets must publish a minimum second data layer — the number of active contracts, payment status, and an independent complaint mechanism for players. Without those three things, any claim of integrity is just a claim.

Every conceded goal begins with a warning number. In esports, the warning usually sits somewhere nobody bothered to record.

The contrarian angle: correlation is not causation, and the analytics industry shares the blame

I have to argue against myself, because that is the rule I set after being mocked for a month before Euro 2026, only for Italy to lift the trophy.

First, the correlation between roster churn and wage delays does not prove causation. Both may stem from a third variable: an owner losing the ability to pay. In that case, roster churn is a symptom, not an indicator.

Second, my sample is small and non-random. I collect data from leagues I can access, and those leagues do not represent global esports. Selection error here may be larger than any effect I find.

Third, and this is what I think the analytics industry needs to hear: we are selling certainty built on incomplete data. A dashboard with thirty indices looks more serious than one with five, but the certainty of a conclusion does not increase with the number of indices. It increases with the quality of the control variables.

The irony is that the "small team beats the giant" narrative that media loves conceals precisely the gap I am trying to measure. When a low-budget team beats a high-budget one, the story told is about character. When the low-budget team then loses ten straight because it cannot pay wages, the story is no longer told. Romanticisation only works in one direction.

Data is not for predicting the future, but for seeing the present clearly. And the present of Southeast Asian esports is a professional labour market operating without books.

Signals for the next cycle

In this transfer window, I am tracking four signals, and I am publishing them here so they can be checked later.

Signal one: the number of transfer announcements stating a specific figure. If the 14.5 percent disclosure rate I recorded rises above 25 percent within twelve months, that is a sign the market is maturing.

Signal two: the number of teams publishing an independent complaint mechanism for players. The current figure I have recorded is none.

Signal three: the patch adaptation lag of the region's highest-paid players. If that group has a lag above average, the market is mispricing and I will have evidence.

Signal four: the emergence of at least one regional contract database, run by a league or a players' association. Without it, any analysis of Southeast Asian esports will remain half an analysis.

I will return to these numbers at the end of the season. Not to prove I was right. But to check whether my system still stands.