The Empty Base: When Sports Analysis Is Built on Data That Doesn't Exist

In November, at a café in Chiang Mai, I was handed a twelve-page document. A handsome cover, a clear table of contents: Overview — Context — Tactical Analysis — Risk Matrix — Overall Assessment. I flipped to the middle. Nine analytical sections. Nine tables. And in every cell, one line repeated: "Insufficient information to assess." The person who sent it to me wasn't annoyed at all. He was proud. "See how clean the layout is? Exactly the nine-dimension frame the desk requires." I read all twelve pages in silence, then asked one question: "So what was the original article about?" He paused. "Not clear. The input extraction came back empty." That was the moment I realised the most frightening thing in this trade: we have just built a machine capable of producing reports that look more professional than the truth they describe. EIGHTEEN YEARS, THREE WAVES In eighteen years of watching sport, I have passed through three waves. The first wave was basic data: time, distance, score. The second was positional data and advanced metrics: PPDA, xG, peak speed, sprint counts. The third — the one unfolding now — is automation: machines that read articles, machines that extract information, machines that assemble reports. The third wave promises something magical. It turns every match into a searchable tactical record. But it carries a structural flaw I only saw clearly while sitting in a commentary box in Kuala Lumpur. In 2026, at the 29th SEA Games, I was the new stadium announcer on the public-address system at the National Stadium in Bukit Jalil. The women's 400m hurdles final. The winner crossed the line in 56.19 seconds. I read it out as 56.89. I also announced her country incorrectly. The jeers rolled down from the stands like rain falling upward. That night I rewound twenty hours of audio — not to apologise again, but to find the pattern in my errors. I discovered that I consistently added roughly half a second to races with large, loud crowds. The cheering made me hear slower. 0.7 seconds is the smallest number that ever taught me the biggest lesson. It taught me that data does not appear by itself. People create it, and people pump into it errors they do not even notice. Ten years later, inside a different system, that error had not disappeared. It had only changed shape. WHEN AN EMPTY CELL IS READ AS A CLEAN BILL The worrying part is not the empty report. The worrying part is its form. A table with column headers, units of measurement and note fields emits a signal of competence all by itself. When you see a "Risk Matrix" with rows labelled "Competitive risk", "Financial risk", "Personnel risk", your brain registers that somebody had a system. Somebody classified. Somebody thought. But if every cell reads "insufficient information", that system is just an empty frame painted over. And that frame carries its own danger: it causes silence to be read as cleanliness. Medicine holds a hard principle: a test that was never performed is not a negative test. In sport, that principle is broken every single day. With no news of unpaid wages, people conclude a club is healthy. With no match-fixing allegation, people conclude a league is clean. With no injury report, people conclude a roster is at full strength. All three conclusions are logically wrong. No data means no data. It is evidence of nothing except its own absence. In esports the error is even more naked. An analysis that cannot identify the game title — whether it is League of Legends, DOTA 2, CS2 or Valorant — renders every conclusion about patches, meta, regional strength and transfer value meaningless. Riot patches every two weeks. Valve patches less often but each change cuts deeper. Tencent runs on a seasonal rhythm. Those three rhythms produce three kinds of players, three kinds of teams, three kinds of crisis. Analysing without knowing the title is fabrication. There is no more accurate word for it. THIRTY PAGES AND THE GATE ON PAGE TWO In 2026, the pandemic closed every stadium. My contract as an announcer for an athletics meet was cancelled. Instead of panicking, I retreated into a study of 58 Bundesliga matches played in empty grounds. The headline result: home win rate fell 12%. But the headline was not what stayed with me. The micro-changes did. Borussia Mönchengladbach cut their pressing volume to 0.78 pressures per minute. The frequency of passes down the flanks rose 17%. I wrote a thirty-page report and sent it to an international journal. Page two of that report was a section very few sports writers produced at the time: "Method and limitations". I stated plainly: 58 matches are not 58 independent observations, because one team appeared eight times. I stated plainly: the pressing metric was hand-coded by me, and the disagreement between two coding passes was 4%. I stated plainly: the control season with crowds was distorted by a compressed fixture list. The report was accepted. But the thing I was proudest of was not the findings — it was the method section. Because the method section is a gate. It stops unsupported conclusions from walking through. When the stadium was empty, I realised: data cannot replace a heartbeat. But I also realised the reverse — a heartbeat cannot replace data. The two need each other, and both need a gate standing between them. BROMELL AND THE HOLE IN THE SPREADSHEET In 2026, at the European Championship, I wrote a tactical column shared more than two thousand times. I dissected how Mancini pushed Bonucci up into midfield to create a three-man screen in defence. That piece was right. But its success made me overconfident. At the Tokyo Olympics, I predicted Trayvon Bromell would win the 100m. His start metrics were the best. His peak speed was the best. My model returned the highest probability. He was eliminated in the semi-final. What I ignored: wind. In the final, the wind shifted direction. And Bromell — who had peaked two months earlier — could no longer hold the stride frequency the old data promised. Bromell arrived as a reminder: every spreadsheet has a hole big enough for a human being to slip through. Since then, every prediction piece I write carries a list called "uncontrolled variables". Wind. Pitch surface. Sleep. Family pressure. A phone call at midnight. The things that never make it into the table. One reader wrote to me: "Your work reads more like a scientific study than a prophecy." I took that as a compliment. 4.8 METRES AND THE VOICE FROM THE DRESSING ROOM In 2026, at the Qatar World Cup, I sat in a studio analysing Morocco's defensive block as a linear system. Average distance between full-back and centre-back: 4.8 metres. A beautiful number — tight, tidy, easy to present. Lineker pushed back on air: the decisive factor was spirit. I pushed back with data. I was right on the numbers. But after the match, a Morocco player said something to me I have carried ever since: "We run for each other, not for the system." 4.8 metres is the distance between two human beings on grass. It cannot measure the distance between two human beings inside the same belief. From then on, every analysis I write includes a section: the voice from the dressing room. Direct quotes. Cross-checked against numbers. So that every figure carries a pulse alongside it. THE PARADOX OF CONFIDENCE The irony is that the sports analysis industry rewards confidence, not honesty. A piece that says "I don't know" will not be shared. A piece that says "if A then B, probability 62%, margin of error plus or minus 8%" gets dismissed as indecisive. A piece that says "this team will definitely win" — even when wrong — gets remembered. I used to think that was the audience's fault. Now I think differently. It is the fault of those of us in the trade, because we taught the audience that certainty is a virtue. The gate I described above is not only needed at the technical layer. It is needed at the ethical layer. An analysis has an obligation to state clearly what it knows and what it does not. If it cannot do that, it is not analysis. It is decoration. I learned to measure time first, and only later learned to measure truth. And the hardest of those three lessons is the third one: learning to say "I don't know yet". CLOSING A 0.7-second error is not the clock's fault — it is the limit of how we frame the question. This season is long. There will be more twelve-page reports with nine empty sections, more risk matrices containing no risk, more conclusions with no data behind them. The problem is not how to make the tables look prettier. The problem is this: when the data cell is empty, do we have the courage to say so — and to endure not being shared?