Esports Data and the Trap of Sourceless Numbers

One November morning, I opened my inbox and found a six-page analysis report. Every field in it was empty. No tournament name, no patch version, no player name. Only the words "insufficient information" repeating like the steady tap of a typewriter that had run out of ink. The sender wrote that the original "had no data to extract." I sat still in front of the screen for a long while, not out of frustration, but because I suddenly realised this is the worst thing that can happen to someone in my profession. The frightening part is not analysing something wrong; it is having nothing to analyse at all. And more frightening still: it is quietly spreading across an entire industry. Esports lives on data. Every match in League of Legends, Dota 2, CS2 or Valorant generates thousands of data points: side win rates, pick-ban rates, gold per minute, damage, objective control time. Platforms such as Oracle's Elixir, gol.gg and Liquipedia collect and publish them almost in real time. The volume is enormous, but volume does not equal quality. The problem is that esports data itself is fragile. A single patch can overturn the meaning of win rates accumulated over months. A champion buffed in today's update turns every statistic from last week into meaningless history. Sample sizes are suspiciously small: a tournament lasting three weeks gives each team fewer than twenty games. That is why I always tell my students the first rule: "Before you trust a number, ask where it was born." Something similar once happened to me in another field. In 2026, while a student in Seoul, I wrote about South Korea's win over Germany at the World Cup, pointing out that the home side's xG was only 1.12 against 2.31 for their opponent. The piece was branded a "betrayal of a historic victory", traffic jumped from 200 to 20,000 visits in three days, and I cried because I was misunderstood. But setting emotion aside, the lesson holds: when I did not clearly state the data source, the measurement scope and the limits of the measurement, my number was no different from a rumour wrapped up neatly. In esports, a number without a source is more dangerous than a wrong number. A wrong number can still be caught. A number without a source drifts through articles, streams and forums, losing a little context at every pass, until it becomes "obvious truth" and nobody remembers who said it first. Take the most familiar example: a team's win rate. It sounds simple, but it depends on which tournament the team played, whether the opponents were strong or weak, which patch it was, and how many games were played. One team can hold a 70% win rate thanks to an easy schedule, while another reaches only 55% but keeps facing title contenders. Placing two numbers side by side without context is a lazy act of analysis, even a deception of the reader. I built an entire process to fight that habit. Every metric I use must answer four questions: who collected it, with what tool, on what sample size, and over what period. If one of the four has no answer, that number does not enter the conclusion. It may appear as a secondary observation, but never as a pillar. This takes time, but experience has taught me that most mistakes in esports analysis do not come from algorithms; they come from input data contaminated at the source. Based on my experience following matches, I once received two datasets from two different providers for the same game, differing by thirty percent on damage. Neither admitted being wrong. Only when I cross-checked against the replay did I learn that one had counted damage to minions. A small detail, but enough to reverse the conclusion about the best player of the match. The "Data Lens" column I run was born from that need. When analysing a transfer, I do not look at goals scored; I look at xG per 90 minutes alongside positional context. Some strikers look weak on the stat sheet only because they were deployed in the wrong role, and conversely, some beautiful numbers come from a whole team serving one player. Here is a counter-intuitive view I want to state plainly. Community excitement is not evidence. A metric shared, commented on and clipped by thousands of people proves only that it is attractive, not that it is correct. Communities are very good at finding holes, but not good at confirming truth. I learned this after my article about Ronaldo at Euro 2026, when I was attacked hard and nearly deleted the piece. What saved me was not agreement, but publishing the entire raw dataset so anyone could check it themselves. At the same time, correlation is not causation. A team winning often while using a certain lineup does not mean the lineup is the cause. Perhaps they simply met weaker opponents during that period. Without isolating variables, we turn a coincidence into a doctrine and then spread it as gospel. The night of Seoul 2026 taught me that the truth can be lonely, but it is never wrong. That empty report, in the end, was not a failure. It was a timely reminder: when there is no trustworthy data, the most honest thing is to say we do not yet know. In an industry always pressured for fast conclusions, verified silence is a luxury. Data does not shout, it whispers — and I have learned to lean in and listen. A question for you next week: the last time you shared an esports number, did you truly know where it came from?