When the Data Sheet Is Empty: Sports Analysis and the Cost of Inventing an Answer

In May 2026, the Bundesliga returned after the pandemic in stadiums without a single spectator. I sat in front of a screen with a table containing exactly two numbers: a home win rate of 43% before the pandemic, and 36% once the ball rolled in empty stadiums. No player names, no possession figures, no heat maps. Just two numbers, and a larger question: is home advantage real, or merely a belief handed down through generations?

I wrote "Home advantage is just an illusion" in two days. It went up on Medium and drew 2,000 reads in 24 hours. A month later, when the Premier League restarted, the home win rate in England rose to 45% — higher than even the pre-pandemic level. I had to sit back down, open every file, and write a correction half again as long as the original piece.

The lesson was not that I was wrong. It was that I had let two hollow numbers fill a conclusion they could not support. And that is a story I meet again every day, at a far larger scale.

Modern sports analysis runs on a paradox. The more data there is, the more people can speak; but the more people speak, the thinner the share of statements that have any basis. Every day, thousands of reports, hundreds of podcasts, dozens of rankings are pushed to market. Every report needs a headline, every podcast needs a take, every ranking needs a placing. That pressure does not produce analysis — it produces answers.

I see this most clearly in the field I follow: esports. A title like League of Legends or Dota 2 shifts its balance with every patch. A single patch can invert the strength order of an entire region within two weeks. The analyst faces pressure to say something immediately — while the data from the new patch is usually only enough to say that there is nothing to say yet.

In football, the same pressure wears a different coat. Every transfer window, a deal worth 100 million euros is announced, and within hours an entire content ecosystem has built the story of how that player will transform the club. People rarely wait to see where he actually touches the ball, which gaps he runs into, and what weaknesses his club exposes once he arrives.

That is why I began to treat "not enough data" as a legitimate answer — even the most valuable answer an analyst can give.

In 2026, I was 25, working as a production assistant for a sports channel in Los Angeles. Before the California Clásico between LA Galaxy and San Jose Earthquakes, I argued straight back at former international Landon Donovan that "winning mentality" was just a fallacy. I cited the first leg's xG: Galaxy created 2.8 xG but lost 0-1, while Earthquakes won on a single moment. He waved it off: "Don't teach me football." The clip spread, and I received 500 comments laced with sexism. I spent three weeks relearning Opta data analysis.

But the deeper lesson was not that I had a number. It was that I used the number as a shield, not as a tool. An xG of 2.8 does not say Galaxy deserved to win — it only says Galaxy created chances in positions the model estimates carry a scoring probability. To understand the match, I had to break xG into each phase: which shot came from inside the box, which from outside, where the opposing keeper stood, whether the defensive line was stretched. An aggregate figure cannot replace a sequence of decisions.

The value of a number lies in knowing what it was counted from, by whom, and what it leaves out. That is the first principle I set for myself, and it is the principle most sports content today violates every day.

In 2026, I applied that principle to the World Cup. I built a model from average squad age, passes into the attacking third, and the breakthrough capacity of the stars. The result: Croatia would reach the final through the trio of Modrić – Rakitić – Kovačić. The post on June 12, 2026 was mocked more than 1,200 times. Many betting accounts told me I "liked to guess wildly". Croatia won three straight knockout matches, beating England 2-1 in the semi-final. After that night, the article was shared 5,000 times.

People laughed at my prediction, but no one laughed at how I recounted every number. The difference between a prediction and a wild guess is this: a prediction can be checked, a wild guess cannot. I did not guess Croatia would reach the final because I liked Croatia. I derived it from three variables anyone could reopen and dispute.

That was also when I understood why esports moves faster than football — because esports is not afraid to be wrong. When an esports team loses a match, they reopen the recording, watch each teamfight, recalculate each resource metric, and publish the mistake. That culture makes esports' learning cycle far shorter than a football world where a coach can keep a wrong system all season with no one forcing him to account for it in numbers.

In esports, the dependence on data is exposed even more clearly. A region once ranked among the leaders can fall after a single patch changes how the mid lane operates. If I write "region X is weakening" without naming a patch, without a pick-ban rate, without a date, then the statement cannot be checked, and therefore cannot be disputed. It is just a sound. At best a headline.

A concrete example of how to verify: if someone says "the new patch is making Asian teams weaker", I need three things. First, the patch number and release date. Second, the win rate of Asian teams before and after the patch, split by region. Third, the pick-ban rate of the champions that were changed, to see whether Asian teams really depend on them. Without one of the three, the statement remains only a hypothesis — and an unverified hypothesis should not be presented as a conclusion.

But that speed has a downside. In January 2026, a source at Chelsea told me they would loan Conor Gallagher to Fulham until the end of the season. Wanting the scoop before the press, I posted "DONE: Gallagher straight to Fulham" before the contract was signed. Gallagher then had to issue a statement that "nothing is agreed", and my source cut contact. The bitter part was that this happened right after I had been the first to report correctly that Jordan Pickford had extended with Everton.

I spent three weeks apologising and publishing a detailed analysis of how I had gone wrong. Since then, I set a rule: never use the words "DONE" or "CONFIRMED" for any information without a signature. The transfer window is where people pay 100 million for a promise and call it faith. But a journalist must not live on other people's faith — she lives on what she can verify.

I have a professional dislike for the loan-with-obligation-to-buy model. In form, it is a way for small clubs to reach big players. In substance, it is an instalment contract disguised as an opportunity. The small club takes the player for a season, pays part of the wages, then is forced to buy outright at a pre-set price — regardless of whether that player fits. They are not buying a player; they are placing a deposit on an obligation. When the season ends, they often have to sell that very player, or another young player, to balance the books. Raising talent and handing it over — that is the operating model of most mid-tier clubs, and it is hidden beneath the language of opportunity.

The same logic applies to load management. People talk about protecting players as a medical standard. But when you look at the schedule, the matches players are rested for usually coincide with fixtures of low commercial value, while they still play in pre-season tours and friendlies. Load management, in many cases, is not rest — it is the reallocation of playing time according to contract value.

And here is what I want to say plainly: when the data sheet is empty, the only honest answer is "not enough information". In deep analysis work, I have received frameworks with every heading filled in: patch analysis, tournament-system analysis, roster analysis, club-finance analysis, risk analysis. But the input data contained not a single information point. No game title, no team, no player, no tournament.

The temptation then is to fill the blanks with plausible-sounding claims. I could write about "meta trends" without naming a game. I could talk about "the financial pressure on small clubs" without naming a single club. Such sentences read very smoothly, and they are entirely worthless. An analysis with no data is not an analysis — it is a writing exercise dressed up in technical terms.

With an empty table, every section must be marked "insufficient information to assess". An analyst who marks that has not failed. It is evidence that the method is working correctly. A framework is only trustworthy when it refuses to draw a conclusion it cannot support.

Here I must argue against myself. The position "if there is not enough data, say nothing" sounds very moral, but it can become an excuse for evasion. For years, I have seen analysts use "we need more data" so they never have to make a prediction at all — and a prediction that is never wrong is also never right. An empty stadium does not make the away team stronger, it only strips the mask off the home team. Empty data is the same: it does not make the analyst weaker, it only strips the mask off those who were fabricating all along.

The real risk is not in saying "I don't know". It is in using "I don't know" as an escape from the duty to make a checkable judgment. I still have to count, still have to put out a number, rather than hide behind caution.

The second blind spot: data never speaks for itself. It is selected, cut, and arranged. When I published the Croatia model in 2026, I chose three variables. Had I chosen three others, the result could have flipped. The objectivity of analysis does not come from having no choices — it comes from disclosing the choices so others can cross-check. A good hot take is not about daring to be wrong, but about daring to be right before the whole world, and daring to let the whole world reopen your spreadsheet.

If there is one thing I want fans to carry away from this piece, it is a question to ask before every report: is the writer giving me a number, or giving me a feeling? A number can be disputed, and precisely for that reason it is trustworthy. A feeling cannot be disputed, and precisely for that reason it is cheap.

Sport will always need people who deliver a verdict fast. But it needs more of those willing to say "I don't have enough data here". In 2026 I stood alone before the whole world. It turned out to be the most valuable position. And I am still counting.