When the Spreadsheet Is Empty: Esports' Data Integrity Problem

I open a file named “esports”. The first row holds the column headers: tournament, date, team, player, metric. From the second row down, every cell is blank. Not a number, not a name, not a timestamp. A spreadsheet created only to prove it has nothing to say. I have sat with spreadsheets for nearly a decade. Long enough to know that an empty spreadsheet is not neutrality. It is a statement. Someone named the columns, built the frame, prepared room for evidence — then put nothing inside. During this transfer window, I have received no small number of reports with exactly that structure: labels everywhere — “deep analysis”, “exclusive data”, “predictive model” — and empty content beneath. What matters is that those reports are still read, still cited, still used as the basis for a decision: a roster slot, a salary, a transfer. I once received a twelve-page report on a young player. A handsome cover, charts, comparison tables, even a section titled “potential prediction model”. I read all of it, then asked the sender a single question: where is the raw data? The answer was “with the aggregator”. I asked further: and the aggregator took it from where? “From sources.” And there it was. Those twelve pages have no root. It is a building erected on air, and the handsome chart on the cover only makes it look sturdier. Every great spreadsheet begins with an empty cell and a question. But not every empty cell is a beginning. Some empty cells are simply the full stop of a process that failed — and people still read them as though they were the answer. CONTEXT: A SEASON IN WHICH DATA STOPPED BEING TRUSTWORTHY In 2026, the Vietnamese esports scene suffered a shock that did not come from a loss, but from the very numbers recorded on the scoreboard. In March 2026, Riot Games announced sanctions against a large number of individuals within the VCS system for match-fixing. Dozens were suspended, several teams were removed from competition, and the organiser had to restructure how the entire league was run. The official figure announced at the time was 32 individuals. For fans, that was a wound to trust. For me, it was a wound to data. The two are different, and I need to be explicit about why. This is the kind of event I call “poisoned data”. Unlike empty data — which does not exist — poisoned data exists but lies. A match has a score, has metrics, has duration, has every column filled in. But the match did not unfold the way the numbers retell it. The danger is that the spreadsheet has no column that says “this match was fixed”. For years, people believed esports had an advantage over traditional sport in that every action is digitised. Every click, every pass, every teamfight leaves a trace in the server log. But a log records only what happened, not why it happened. When motive is bent, the log is still clean. Error does not lie — it merely whispers what we are not yet large enough to hear. I have followed the VCS since its earliest seasons, when the league was still seen as one of the most promising regions in Southeast Asia. In 2026, a Vietnamese team made the whole world say its name after a match on the international stage. Then, in the seasons that followed, the gap with the major regions grew ever clearer. Looking back, I ask myself how much of that gap was real strength, and how much was the consequence of matches whose data was never trustworthy. That question can be answered, if we have a clean enough dataset. THREE LAYERS OF BROKEN DATA I divide broken data in esports into three layers, in ascending order of danger. The first layer, poisoned data. This is the easiest to spot because it attaches to a legal or disciplinary event. When a match is fixed, all the data born from it — win rates, player metrics, transfer valuations — becomes counterfeit. But the problem does not stop at that match. A team’s data over a season is an aggregation of many matches. If three out of twenty are bent, that team’s win rate is no longer a measure of ability, but a mixture of ability and guilt. And because nobody labels each match, we are forced either to discard the whole season or to live with the contamination. After VCS 2026, many historical datasets from this region became worthless. One cannot compare a player from that period with a player from a later one, because the yardstick was broken at the root. This is a cost few count when they speak of match-fixing. They speak of honour, of sanctions, of fan trust. Few speak of how an entire database — the thing used to scout, to value, to build strategy — was silently disabled. Professionally, this feeds directly into the transfer market. The transfer market is where emotion is defeated by probability — that is a principle I have always believed. But the principle holds only when probability is computed from real data. When data is poisoned, probability becomes a shield to justify decisions already made. A club can buy a player at a high price, then invoke his “impressive metrics” from a questioned season, only to discover months later that those metrics did not convert into real ability. The second layer, empty data. This is the least noticed layer, because it attaches to no scandal. It is simply numbers invoked without a source. During the transfer window, I read hundreds of pieces containing sentences like: “This player reaches 0.28 expected assists per 90 minutes.” It sounds very scientific. But how many state clearly which league, which season, what sample size? Very few. A number without provenance is not a number. It is a rumour with a decimal point attached. I have a rule when reading any report: if the source cannot be verified, I mark the entire report as “unverified”, however well the rest is written. Because a wrong number in the first line drags every conclusion in the lines that follow. The label “deep analysis” does not compensate for an empty data source. There is a variant of this layer I call “circular citation”. One article cites a report. The report cites another article. That article cites the first article again. All three lean on one another, none holding raw data. Technically, this is a closed system with no anchor to reality. But formally, it looks highly credible, because many sources say the same thing. A crowd does not create truth; it only creates the feeling of truth. The transfer window is when the empty-data layer shows most clearly. Noise drowns out signal. Every day brings dozens of rumours, each fitted with a number to raise its credibility. A transfer fee is stated, nobody knows what it rests on. A salary is mentioned, nobody can verify the source. Fans drown in rumour, and what they need is not more rumour but a reliability filter. That filter is not hard to build: simply ask of any item three questions — who says it, on what basis, and where can it be checked. There is another technical problem I want to spend a few lines on. Esports often has small samples. A tournament lasts a few weeks; a team plays a few dozen matches a season. That sounds adequate, but once you split by patch, by opponent, by role, each data cell holds only a handful of observations. At that sample size, a single win can shift the whole statistical picture. That is why I always say: one match is noise; one season is signal. But even a season, if sliced too finely, can turn back into noise. The third layer, misread data. This is the most dangerous layer, because the data here is entirely true; only the reading is wrong. And the chief culprit is usually the patch. There is a variable that is always present before every final yet rarely named: the patch. At Worlds 2026, T1 beat Weibo Gaming 3-0 in the final in Seoul on 19 November. At Worlds 2026, T1 won again, beating BLG 3-2 at The O2, London, on 2 November — the team’s fifth title. After each such run, the analytical world rushes to explain. Some say it was Faker’s mentality. Some say it was coordination. Others say it was the coaching staff’s drafting. But the foundation of every such explanation is usually skipped. A champion buffed, an item nerfed, a match tempo shifted — all of it happened before the final began. The champion team is usually the one that adapted best to the competitive version, not necessarily the strongest on paper. But adaptability is a hard skill to measure, while the patch is an easy variable to see. People choose to measure the hard thing with the easy thing, and the result is credit assigned wrongly. The case of Gen.G and Chovy is worth thinking about. Gen.G dominated the LCK across many seasons, Chovy sat perpetually among the players with the highest individual metrics in the league, yet the Worlds title remains untouchable for them. This is usually explained by psychology. I do not deny psychology. But every time Worlds arrives, the patch shifts in a direction different from the domestic league, and teams whose playstyle was built for the old version tend to pay. Part of that story lies in the data; it is just that most readers do not separate it out. A player with a high win rate at one tournament may simply be someone playing exactly the champions the patch favours. That win rate is then compared with another player in another tournament, on another patch. The comparison is meaningless from the first step, yet it is still printed, still shared, still used to conclude who is better. I once built a model to separate these two variables. I took one player’s data across multiple patches, recording the times his signature champion was buffed and the times it was nerfed. The result was not entirely clear — but it was enough for me to realise that most of the fluctuation in a player’s win rate comes from changes he does not control. That is something a simple ranking never tells you. In Southeast Asia, GAM Esports of Levi is the team that has appeared at many World Championships. Each time this team goes deep, a wave of analysis about “Vietnamese mentality” follows. I do not deny mentality. But I want to point at something else: most of the results achieved by Southeast Asian teams on the international stage attach to periods when the patch opened space for early aggression — a style these teams have always played well. When the patch changes, the gap widens again. If we call the favourable period real strength and the unfavourable period decline, we have misread the data at both ends. A shock is only data that history has not yet had time to name. The shock at Worlds 2026, when an underrated team eliminated the reigning champion, was also a form of misread data. At the time, everyone spoke of spirit and mentality. Very few spoke of how that year’s patch neutralised the strong team’s control style and opened space for high-pressing teams. The data was there. Nobody read it. THE CONTRARIAN ANGLE: THE TRAP OF PRESENCE This is a paradox it took me years to name. We react very differently to absent data and present data. When a cell is empty, we are cautious. When a cell holds a number, we believe — regardless of where that number came from. The presence of a number is mistaken for its value. That is why the second layer — empty data in the clothing of real data — is the hardest of all to detect. It causes no scandal. It is not suspended. It simply drifts quietly through articles, transfer bulletins, rankings. And little by little, it erodes an entire community’s ability to distinguish signal from noise. I used to think esports’ biggest problem was bad data. I was wrong. The biggest problem is unexamined data. A system can live with bad data if it has a detection mechanism. But it cannot live with a process that produces empty spreadsheets and pastes the label “deep analysis” on top — because at that point, the whole system is deceiving itself in an organised fashion. There is a point I want to state plainly, even if it is hard to hear. In esports, most “analysis” is written not to find the truth, but to rationalise a belief already held. Data is selected to fit the conclusion, rather than the conclusion drawn from the data. When a team wins, people look for numbers to explain the win. When a team loses, people look for numbers to explain the loss. And the patch — the variable that actually decides — is pushed into the footnote. Another baseline deserves mention: data cannot measure what never happened. No column records the moment a player hesitated for a second before a teamfight. No metric measures the pressure of a match that decides qualification. A spreadsheet sees only what has already surfaced. Most of a season’s story lies in what was never recorded. That is why I always leave a gap in every model of mine — a deliberately empty cell, to remind me there are things I do not know. When the stands were empty, I heard data speak for the first time. In 2026, when leagues had to play without spectators, I had a rare chance to separate the crowd’s influence from match outcomes. Home-team win rates fell, average goals fell — numbers that would normally drown in the roar. That silence taught me that data speaks only when enough noise is removed. And most esports data today is still far too noisy to say anything certain. SIGNALS FOR THE NEXT CYCLE So how should esports be read properly next season? I propose three rules, all of them applicable immediately. First, the provenance rule. Every number must come with three things: data source, time range, and sample size. Missing one of the three, the number is downgraded to rumour. It sounds severe, but this is the minimum standard any serious analytical field must hold. Second, the version rule. Before comparing two players, two teams, or two seasons, ask whether they share the same competitive patch. If not, the comparison holds only reference value. Separating the patch’s influence from a team’s true ability is the hardest exercise, and also the one fewest attempt. Third, the admission rule. A model is permitted to say “I do not know”. A spreadsheet is permitted to be empty. An analysis is permitted to end on an open question. Humility is the condition for analysis to be trustworthy. As for the VCS, I believe the signal to watch in the period ahead is not any single team’s results, but the speed at which data integrity is rebuilt. Once there is a clean, continuous dataset recorded to the same standard across multiple seasons, only then can one begin to speak seriously about this region’s true strength. Until then, every comparison is merely a game of labels. What the world calls a miracle, my spreadsheet saw from winter. But a miracle is visible only when the spreadsheet is clean enough to look. A poisoned spreadsheet will see nothing. An empty spreadsheet will see nothing. Only an honest spreadsheet can see what is about to happen. As for me, I will still open my files every morning. Some will be full of data. Some will be empty. And I have learned that sometimes, being honest with an empty cell is worth more than a spreadsheet stuffed with falsehoods. Every number is one meditation; every season, one awakening.