From the pitch to the esports arena: The art of interrogating numbers that lie

On October 21, 2026, Huddersfield Town beat Manchester United 1-0 at John Smith's Stadium. The scoreboard told one story; the data told another. Manchester United generated 1.82 expected goals; Huddersfield only 0.35. The visitors had more possession, more shots, and created chances any model would have to acknowledge. Yet the three points stayed with Huddersfield. I watched the tape of that match for a week, not to find the goal, but to find the reason. The reason turned up somewhere no major outlet bothered to count: 27 tackles in front of the box. Nothing flashy, just blocks, interceptions, and interruptions at exactly the right moment. xG measures chances, but it does not measure ferocity. That year I was eighteen, living in a college dorm in Chicago, writing a football blog for myself. The Huddersfield moment shaped how I have looked at every number since: the match in which xG lied is the match in which every number must be interrogated from scratch. Years later I moved from the pitch to the esports arena. From football to esports, from xG to role-based metrics, from distance covered to the intensity of ball handling. The lesson never changed: data does not speak the truth on its own. Data only echoes what the person reading it wants to hear. And there is a kind of truth more uncomfortable than a lying number — the truth of a number that refuses to speak. The esports industry went through a data revolution later than football, but many times faster. In Europe, it took until the late 2000s for metrics like xG to creep into football coverage. In esports, within roughly the last five years, every major competition — from League of Legends to Dota 2, from Counter-Strike to Valorant — has gained its own data layer: post-game stat sheets, role metrics, composition win rates, and models running behind the scenes. What caught my attention is that most methods esports now uses are borrowed from football. People take ball possession to talk about map control. They take expected goals to talk about high-probability team fights. They take transfer value to price a player. Every time a method is borrowed, a layer of context gets stripped away. That is exactly where numbers begin to lie. I once sat in an analytics meeting at an esports organization. On the screen was a player's stat sheet with a very high kill-to-death ratio. The room nodded. But when I asked how many difficult situations that player had handled, and how many times teammates had repositioned just so he could keep his clean number, the room went silent. A beautiful number can be the mark of a star, or the mark of someone benefiting from a system. A number never incriminates itself. In esports, I hear the echo of football before the data era. Once, people judged players by goals scored. Now, people judge players by kills. Both are surface numbers, easy to measure, easy to show off, and easy to deceive with. Anatomy of a number Any esports metric has three layers. The first is the raw number: kills, deaths, gold earned, damage dealt, vision created. The second is context: who the opponent was, what the composition looked like, how long the game lasted, and the role within the tactical system. The third is meaning: what that number says about the team's future ability to win. Most fans read only layer one. Most amateur analysts stop at layer two. Only a few bother to reach layer three. Take a team-based title like League of Legends. A mid laner with high gold per minute might be a superstar, but he might also be the result of the whole team funneling resources to him. Put two numbers side by side without context and it is easy to draw the wrong conclusion about a player's true value. I have seen reports praise a player simply because his gold-per-minute topped the league, ignoring that his team won thanks to a bottom laner quietly absorbing pressure all season. In a tactical shooter like Counter-Strike or Valorant, the story is subtler still. Average damage per round and the percentage of rounds with at least one kill are two very different things. A player may rack up damage mostly in already-decided rounds. Another may deal less damage but open up the rounds that decide the map. If you only look at the final scoreboard, you see the first player in lights. If you watch the tape, you see the second player holding the rhythm for the whole team. Here, the Huddersfield lesson returns. Huddersfield's 27 tackles never appeared on the scoreboard. They did not appear in xG either. They appeared only when someone took the time to sit down and count. Data is never in a hurry; it waits until you are clear-headed enough to ask the right question. Most esports data lives at layer two, but few bother to descend to layer three. There is another trap I run into constantly in esports reports: confusing correlation with causation. A team often wins when it controls vision, but that does not mean vision control is the cause of victory. Both may be the result of something else — say, the map-reading and decision-making of the in-game leader. Esports sample sizes are small, noisy, and shaped by the game version. A metric can be strong this season and weaken the next simply because the publisher changed a mechanic. Without repeated samples and clear causal evidence, any claim about cause is just a guess dressed up in numbers. I once watched a team pour its entire analytics budget into a single metric because it looked convincing in last season's data. When the new patch landed, the metric became meaningless, and the team had no fallback. That is the price of trusting a number without understanding the context that produced it. Heat maps — the new divination There is one tool I grow more wary of the more I use it: the heat map. A heat map shows where a player spends the most time on the map, and it looks very scientific. But a heat map hides a player's real role within the tactical system. A player who frequently occupies one zone may be there because that is his assignment, not because it is where he is best. A player roaming the whole map may be a lost wanderer, or may be the link between the top and bottom lanes. A heat map cannot tell the two apart. I call the heat map the new divination of the analytics trade. It offers a feeling of precision without demanding understanding. People look at a red-hot streak on a map and believe they understand a player. But to understand a player, you need to know what he does in those positions, at what times, and to serve what goal. Without those three things, a heat map is just a pretty picture without a caption. When data refuses to speak Now I want to describe a different kind of failure, the one I consider the noblest in this trade. It is the moment an analyst looks at the data and says plainly: there is not enough here to conclude. I once received a request to analyze a tournament whose input was empty. No event name, no teams, no players, no game version. There was a single label: esports. In that situation, the right move is not to invent a plausible-sounding analysis. The right move is to stop and say it cannot be assessed. There is an iron rule in my work: missing data is not zero data. If there are no signs of financial risk, that does not mean there is no financial risk. If there are no signs of a competitive-integrity violation, that does not mean things are clean. The silence of data is a gap, not a statement. Confusing the two is one of the most dangerous mistakes in sports analytics, because it turns ignorance into a conclusion. When an analysis is built on an empty input, the only possible output is fabrication. And fabrication, once dressed in numbers, becomes very hard to detect. It looks professional. It has tables. It has decisive conclusions. But underneath, it has nothing. That is why I always tell young people who want to work in sports data: the most important skill is not knowing how to run a model, but knowing when to stay silent. A good analyst can say many things from little data. An honest analyst knows that sometimes the truest answer is a question turned back on the asker. The transfer market — a mirror of fear If there is one place where data and emotion fight hardest, it is the transfer market. The transfer market is just a mirror reflecting the fear of managers. A club buys a star not only because he is good, but because of the fear of being left behind. A club keeps a player not only because he is needed, but because of the fear of losing what it already has. The story of Sofyan Amrabat is one I remember clearly. After his brilliant performance at the 2026 World Cup for his national team, I prepared a long analysis recommending that a club spend money to bring him in. The analysis rested on his ball-recovery numbers and his reading of the game. But the decision-maker rejected it with a very different argument: Amrabat has no commercial value, nobody buys his shirt. By the summer of 2026, Amrabat had moved to Manchester United on loan, and my analysis began circulating through professional offices. The lesson is not whether I was right or wrong. The lesson is that a technically correct analysis can still fail if it is not presented in the language the decision-maker craves. Managers do not buy data. Managers buy peace of mind. If you hand them a beautiful number without tying it to how it helps them make money or protect their reputation, that number gets ignored. In esports, the same thing is happening. Esports organizations pour money into players with big followings, sometimes more than they pour into players with better competitive metrics. The esports transfer market, like the football transfer market, does not reward accuracy. It rewards storytelling. The blind spot of precision There is a paradox I want to put on the table: the more precise you are with numbers, the easier it is to go blind to people. When I began working with positional data, I discovered that the best metrics for measuring a team are sometimes not the ones that get published. I remember a study I once did on the effect of playing in empty stadiums. During the pandemic lockdown, I pulled data from dozens of matches and compared them with the period before. The result surprised me: home teams' win rate dropped significantly, and draw rates spiked. Home advantage, long treated as a constant of football, turned out to depend heavily on the crowd. That finding taught me something: sometimes the most important variable is the one nobody bothers to measure. The crowd is not a number in any tactical model, yet it changes the outcome of a whole season. In esports, the forgotten variable might be the noise of the arena, the latency of the connection, or the psychology of a young player competing for the first time in front of thousands. No model measures that fully. And no model should pretend that it does. What the scoreboard does not tell Back to Huddersfield. What I learned from that match was not that xG is useless. What I learned was that xG is useful only when placed beside other things. The road to a final is not in the feet, but in the distance they are willing to run. At the 2026 World Cup, I tracked dozens of matches and noticed that a team the press called old and slow had one of the highest average distances covered per match in the tournament. They did not win with flashy numbers. They won with endurance in extra time and the ability to hold their rhythm when opponents had faded. Those numbers never appeared on the scoreboard, but they decided who advanced. In esports, matches stretched across multiple games operate on similar logic. A team that wins the first game on a moment of brilliance can lose the whole series if its mental stamina and adaptability are weaker. Data cannot measure fatigue. But data can measure the gradual slowdown of decision-making over time. That is the kind of data I consider most valuable, and also the kind fewest people bother to mine. The contrarian view: number addiction The sports data industry is suffering from an illness I call number addiction. People prefer a number to a person. They trust a table more than a story. They use data to attack players, to belittle fans, to prove they understand more than others. That is when precision turns into arrogance. But the one thing data can never do is replace humility. The best analysis must begin by admitting it might be wrong. It must leave room for what cannot be measured. It must distinguish between what it knows and what it is merely guessing. There is a truth people in my line of work rarely say out loud: most esports conclusions are built on small samples. A tournament has a few dozen matches. A player has a few hundred games. Compared with football, where thousands of matches are played each season, esports data is far thinner. That means esports analysts should be more humble, not more confident. In practice, the opposite is true: because there is little data, people are quicker to assert strongly, to compensate for their own insecurity. I do not believe in luck, but I believe in the probability of forgotten shots. In esports, there are forgotten plays too: movements that produce no kill, blocks that never make the scoreboard, calls made at the right moment in the chaos. These are the things data has not yet reached, and perhaps never fully will. Signals for the next cycle So what should fans do with all this? The first thing, I think, is to learn to doubt responsibly. Do not trust a number just because it exists. Ask in what conditions it was measured, by whom, and to answer what question. The second is to accept that some questions today's data cannot answer. Maturity in analysis is not having many answers, but knowing which questions remain open. The third, and perhaps most important, is to keep the number in its place. Data is a guide, not a judge. It helps us understand a match more clearly, but it never replaces the unpredictable complexity of sport. Every match is a confession, and my job is to read between the lines of code. The next trend in esports data will not lie in collecting more metrics, but in understanding the context of the metrics we already have. The organizations that learn to place numbers in context, tie them to human decisions, and know when to stay silent will lead over the next few years. Those that merely chase flashy tables, uncaptioned heat maps, and models built on empty inputs will soon find they are doing divination in the language of science. Data will not save a team. But someone who reads data with clarity and humility can help a team avoid its most expensive mistakes. And sometimes, in an industry where everyone wants to hear a decisive number, an analyst's greatest value is the courage to say: I do not yet know enough.