The Empty Analysis File and the World Cup Theorem: How Sports Writing Tests Its Own Evidence

In November 2026, in Turin, I filed my analysis of the second leg of the World Cup play-off between Italy and Sweden. On 13 November 2026, at San Siro, the match ended 0-0. Sweden advanced on the strength of a 1-0 win in the first leg on 10 November 2026 in Solna, and Italy missed a World Cup finals for the first time in 60 years. In that piece I showed how head coach Gian Piero Ventura's 4-2-4 left dead space between the lines: the midfield isolated from the back line, the wide players not tracking back deep enough, and Sweden's midfield needing only to hold position to cut every vertical pass. My editor dismissed it in one line: "Girls write tactics for decoration." I spent 240 minutes re-watching the tape, drew 14 pressing maps, logged the minute of every phase, and filed again with the data attached. The piece ran once he had no reason left to refuse it. Since that night I have written by a single rule: no data, no argument. Recently that rule was tested somewhere few would expect: a deep-analysis file returned by a text-processing system. The title field read N/A. The source field read N/A. The list of information points was empty. The entity section was left blank. What stood out was how the system responded: it did not fill the gaps with plausible-sounding sentences. It rated its own information value at one star out of five across all four categories, flagged information-integrity risk as the single largest risk, and asked for the extraction step to be re-run before anyone used the output. In this profession, a document like that is worth more than ten full pages of prose. A serious piece of analysis does not begin with a judgement. It begins with decomposition: the original article, the source, the timestamp, the list of information points, the named entities. Only once that layer stands does the second layer get to run: nine analytical dimensions, from technical, tactical, personnel, competitive landscape, regulation, transfer market, risk profile, public narrative, through to the transmission chain of an entire industry. The conclusion appears last, and it must answer one question only: which piece of evidence is holding it up? There are 22 players on the pitch, but the match is really played between two brains. I learned to see it that way from two different environments. On the race track, every car feeds hundreds of data channels per second to the pit wall, but the strategist does not read raw data. He reads the gap, the tyre wear, the pit window, and only then decides. On the pitch, the raw data is passes, ball recoveries, distances between the lines. Both places share one rule: data does not decide for anyone, but without data every decision is a guess. That is why I used two pandemic years to rebuild the Atalanta data set under Gian Piero Gasperini. In the 2026-20 Serie A season that club scored 98 goals, a figure no side outside the big spenders had produced. When I separated each goal by transition type, the pattern was unmistakable: most goals came after three to five passes from the moment the ball was won in the opponent's half, and the finishing point usually sat in the inside channel where the opposing full-back had just lost his position. Watching tape alone, I would have written that Atalanta "attack with total football". That sentence means nothing. What is worth writing is the mechanism: they force the opponent to pass into a zone where two players are already waiting, then counter within four seconds. Four seconds. A unit of time so small that without logging the minute of every phase, I would never have found it. Based on my own experience tracking Serie A matches and the footage I tag myself, most analytical errors do not come from misreading the numbers. They come from missing a detail that small. The technical dimension of an analysis works in the same way. For a race, it is a question about an upgrade package: does it add downforce, or does it simply widen the tyre operating window? For a football team, it is a structural question: can the midfield hold a distance under 15 metres from the other two lines across the first half? When the data is empty, neither question has an answer. A piece that dares to print "insufficient data" is more honest than a piece of conjecture with a chart attached. In sports data I distinguish two kinds. Descriptive data tells you what happened: possession share, shots, touches in the box. Decisive data tells you why it happened: the position of the third player in a counter-attack, the moment a line was pulled out of shape, the time between winning the ball and playing the final pass. The first kind fills a page very quickly. Only the second kind produces an argument. The tactical dimension demands something else: timestamps. On the track, that is the pit window, the call to extend a stint or stop early for the undercut. On the pitch, that is minute 60, when a coach changes structure or makes a substitution to close the gap between the lines. Spain's 3-3 draw with Portugal on 15 June 2026 in Sochi is the example I still use with interns. Cristiano Ronaldo scored a hat-trick, but the data worth discussing sits on the other side: Isco kept moving into the space between Portugal's midfield and defensive lines, forcing a centre-back to step out, which opened the pass into the channel behind. On my first draft, the desk cut the piece in half because "nobody reads detail like that". I learned to put the main argument in the opening and compress the evidence to the shortest form that still allows verification. The personnel dimension is the easiest to be fooled by. A driver winning one race proves little if his team-mate is running a different strategy. A striker scoring ten in ten is only credible once you know the quality of the chances he was given. Without an internal benchmark, every individual judgement sits at the level of speculation. The first question in every file of mine is always: what standard is this person being measured against? The competitive-landscape dimension sets the ceiling for every conclusion. A team at the start of a regulation cycle behaves completely differently from a team at the end of one. A team sitting close to a spending threshold will prioritise efficient upgrades over glamorous ones. Skip that context and analysis becomes a league table with footnotes. The regulatory and governance dimension is the one I check before publishing, because it decides what is allowed to happen. A technical detail in the grey area of the rulebook can turn a win into a penalty a few weeks later. A mid-season format change can erase an advantage a team spent months building. Sports forecasting without reading the rules is blind forecasting. The transfer market dimension connects analysis to real resources. Every empty seat is a variable, every extension clause an assumption about the future. Here information often arrives from sources that cannot be verified, so I apply a rule of my own: use the information only when the motive of the person releasing it can be identified. A leak can be right about the event and wrong about the intent. Verification is a chain, not a step. I rank sources in three tiers. The primary tier holds official documents, official timing data and records. The middle tier holds reporters who were present and industry specialists. The inference tier holds models and the statistics I build myself. Every conclusion must state which tier it stands on, because an inference-tier claim must never be presented in the voice of a primary-tier fact. The risk-profile dimension is where I keep one question for myself: where will this team crack first? Systemic risk usually comes from somewhere nobody watches: a technical detail never tested at high temperature, a team-mate relationship going cold, a sponsorship contract nearing expiry. For a football team, systemic risk lives in squad depth: the first XI is strong enough, but one injury in a spine position knocks the whole structure out of alignment. The public-narrative dimension is the easiest to manipulate. In 2026, when football returned to empty stadiums, I rebuilt data from 120 matches to test a simple hypothesis: with no crowd, how much home pressure is left? The result showed home teams losing about 15 per cent of the pressure they exert on opponents. The piece was widely shared, and what I remember most is how many people read it: they did not ask about the method, they asked whether the number confirmed what they already believed. An empty stadium is not an anomaly. An empty stadium is an operating theatre. In an operating theatre you see things that the noise used to hide. The industry transmission dimension closes the chain. A decision at the technical level flows down to the commercial level, then to media, then to derivative markets. A new contract flows back up and changes the structure of the team. Every new contract is a hypothesis. The match is the experiment. The biggest problem in analysis today is that models always return an answer. A system designed never to return a blank cell will always find a reply, even when the input is a few lines of headline. And because that reply is fluent, grammatical and illustrated with numbers, readers have no way to tell it apart from a conclusion built on real data. The risk to this industry does not lie in bad data. It lies in empty data filled with plausible prose. I also have to argue against a reflex of my own. The rule "no data, no argument" can easily become a shield: waiting for enough data to speak, and eventually saying nothing. A piece that never commits to a position is just a long note. The grey zone is not where the light is missing. It is where football is most real. A good writer has to step into it, state the confidence level of each inference, and pre-write the alternative scenario in case the first assumption collapses. Alternative scenarios are how I defend myself. Before stating any prediction, I draft two branches: what happens if this team holds its structure until minute 70, and what happens if it loses its midfield spine earlier. When both branches lead to the same conclusion, that conclusion is worth publishing. When only one branch looks good, I am telling a story rather than analysing. I do not believe in titles. I believe in the system that operates to produce titles. Readers can also test a piece of analysis with a few simple questions. Does the conclusion carry timestamps and specific numbers? Is the source named, with a date, or is it only "a source says"? Does the author state what would prove them wrong? A piece that will not name its falsifying condition is usually built to protect a pre-existing view. In a big-tournament season, emotional pressure compresses reason. Fans follow with flags and with stories, and that is entirely reasonable. The analyst's job is to hold the rest: to say clearly what happened on the pitch, what is only inference, and what cannot yet be concluded. My World Cup theorem does not predict the champion. It predicts who collapses first. The most valuable question remains the one about breaking points: who is accumulating tactical debt, who is hiding a crack behind a run of good results, who can hold structure when the pressure doubles. The next round will answer. And the data sheet will speak first.