The Zero-Input Trap: Why Football Data Analysis Collapses on a Faulty Foundation

2:40 a.m. In my London flat, under the blue light of a laptop, I am scrolling through an analysis that landed on my desk. The formatting is flawless — tidy headings, clean tables, every cell filled where it should be. But the first-layer input was substantively empty: no title, no body, just blank cells and one "N/A" after another. I stopped the cursor. I have watched football for more than four decades and chased numbers for a long stretch of that time, and in that moment I understood — the biggest risk in football analysis is not a wrong model, it is a zero input. A model never says "I don't know." It always produces a number, even when the input is empty. An xG model will print 0.00 on an empty dataset, and we will accept it as "neutral." Yet where there is no information, there is no neutrality; there is only ignorance, dressed in the costume of a number. This piece tries to take that costume off — in the noise of a transfer window, where rumour and truth live at the same rate, I want to show why analysis built on a zero input can cost a club an entire season. We are inside a transfer window now. What sells best in this market is not information — it is confidence. A tweet, a "close source," a photograph of someone standing in an airport: those three things are enough to build a club's future. I watched the transfer market like a monastery ledger: quiet, exact, unforgiving. A ledger never records a rumour; a ledger records the fee, the wages, the release clause, the agent's commission, and the years left on the contract. That is where this piece must begin — on paper, not in a headline. My method is simple but patient. First I fix the question — "will this forward score 20 goals in the Premier League," or "will this centre-back survive a low block." Then I pull the historical data: open-play xG per 90, the share of shots inside the box, progressive passes, PPDA, and set-piece xG. Then comes the most neglected step — league adaptation. Serie A's tempo is not the Premier League's tempo; La Liga's technical space is not the Premier League's physical duel. An analysis that skips this step is not using data — it is misreading data. June 2026. Liverpool signed Mohamed Salah for £36.9m. I had just begun a numbers-first column for a London outlet. For 72 hours I pulled every Serie A shot from Roma's 2026-17 season. Salah's open-play xG per 90 was 0.52, and 68 percent of his shots came from inside the box. The numbers told a clear story: this was not a winger, this was a 25-goal forward. I wrote a 2,000-word piece arguing that Liverpool had bought a goal machine, not a delivery man. He scored 32 Premier League goals that season. My model beat the eye test. But a subtle lesson hides here, one I have underplayed for years. Salah's success did not come from his xG alone — it came from a change of role. At Roma he was a winger pinned to the touchline; at Liverpool, Klopp pushed him inside the box. My model was measuring the right thing, but the real cause — the system and the manager's trust — sat outside the model. A correct prediction can be correct for the wrong reason. Miss that distinction and the analyst grows confident, and a confident analyst will make his next call on a zero input with the same confidence. July 2026. Barcelona signed Robert Lewandowski for €45m. He was 33. Many called it a late, misguided investment. I built a La Liga adaptation model. His 2026-22 Bundesliga season: 35 goals, 30.5 xG, 4.1 shots per 90. The numbers said the goals were still there; the warning sat in his pressing — his PPDA involvement had fallen 12 percent. I projected 25+ La Liga goals, with a caveat about pressing decline. He scored 23 league goals. The model was almost exact, but the gap between almost and exact taught me something: without pressing data, a goal count is half a story. July 2026. Spain lost their Euro 2026 semi-final on penalties. Everyone wrote about the missed penalties. I ignored them. I pulled the numbers of 18-year-old Pedri: 92 percent pass accuracy, 7.3 progressive passes per 90, 0.14 xG. The market saw a teenager; I saw a midfield metronome. From that discovery came the "Young Core Index" — a plan to track Pedri, Bellingham and Musiala for 12 months. I told my team: "We don't cover matches; we cover the next five years." July 2026. Before the Russia World Cup final, France against Croatia. I built a PPDA and set-piece xG model. Croatia had played three consecutive matches into extra time — 90 extra minutes. Their PPDA had drifted from 8.4 to 12.1. France's PPDA was 9.8, and their tournament set-piece xG was 3.2. I told my editor France would win by two goals. France won 4-2. France's set-piece xG had already lifted the trophy in my model before the final whistle. For me, set-pieces are a ledger in which the season's silverware is written before the season starts. If a team generates eight to ten goals a season from corners, and those come from a repeatable routine — fixed delivery zones, fixed second-ball patterns — that is not luck, that is structure. I measure the repetition rate of that routine. Where repetition is absent, my model does not lift the trophy. In France's 2026 case the repetition was there; that is why I was confident. But trusting set-piece determinism means forgetting that conversion rates on corners and free-kicks are unstable in small samples — I treat variance as a term, not as luck to be waved away. June 2026. The Premier League's Project Restart began, with empty stadiums. I pulled the data from the first 40 matches behind closed doors. The home win rate fell from 45.2 percent to 30.0 percent. Home teams' PPDA worsened by 1.7; their xG differential dropped from +0.24 to -0.11. I wrote "The Empty Stadium Effect." My argument: crowd noise is a tactical variable, not merely atmosphere. When the stadiums emptied, my home-advantage variable quietly died. That discovery changed every post-match analysis I write — I no longer read a match without crowd context. League adaptation is the most neglected layer of my model. A forward may be generating 0.7 xG per 90 in the Eredivisie, but the height of the Premier League's defensive line and the intensity of its duels are different. I measure three things: average shot distance, the number of touches inside the box, and the share of shots that break down under opponent pressure. Without those three, I do not trust any league-to-league transfer valuation. Another input everyone avoids — medical data. A player's injury history, his minute-load, his recovery curve. In a transfer window this almost never makes a headline, yet it decides whether a €45m investment is really a €20m one. An analysis that omits this layer of input is building a full decision on a half-truth. A pattern emerges from these experiences, and it is the centre of this piece. Behind every successful prediction of mine there was always one thing: clean, verifiable, complete input. In Salah's case it was the full shot data from Roma; in Lewandowski's, four consistent Bundesliga seasons; in Pedri's, tracking of every pass; in France's, every set-piece of the tournament; in the empty-stadium case, every record from 40 matches. In none of them was the input blank. I have seen the opposite, too. An analysis that starts from a zero input will produce a tidy grid, a handsome chart, a confident conclusion — and everyone believes it, because the format is correct. This is the modern form of "garbage in, garbage out," but more cunning: the output does not look like garbage, it looks like evidence. That empty analysis on my desk was its living example — no title, no body, yet pretty tables all around. Anyone who assumes those blank cells are "neutral" and decides on them is firing arrows in the dark. Now to the most uncomfortable part — the trap of correlation and cause. When a model fails, we assume the data was bad. But the problem runs deeper: a model can be wrong even when its result is right. Salah scored 32 goals; my model said 25. The number matched, but did the cause match? No. The real cause was Klopp's system, a variable outside my model. Had I not asked myself "why" that day, the next time I would have made the wrong call with perfect precision. In a transfer window this trap cuts sharper. A club buys a forward, he scores in three straight games, and we write "successful signing." Yet across those three games his xG was 0.4 — meaning he stands on luck rather than elite finishing. Small samples are football analysis's silent killer. With set-piece xG it is even clearer: a corner routine may deliver two goals in two games, then nothing across ten. Without measuring variance, we cannot tell a tactic from a coincidence. Another trap — model worship. If xG appears in the headline, we assume the analysis matters. But xG is itself a model with its own limits. Which data built it, which league calibrated it, which role the player occupies — without knowing these, xG is just a decimal. I sometimes caution my former self: let the Salah xG success not blind me. One experience cannot explain every position, every league, every era. A striker's xG is not a centre-back's progressive passes; Serie A data is not Premier League data. I have set a falsification test for my own model: if a prediction comes true but I cannot explain why it came true, I do not count it as a "success." Because an unexplained success sows the seeds of the next failure. In Salah's case I later found the explanation — the role change. Where I could not find one, I rebuilt the model from zero rather than decorating it. The real integrity question, then, is not about a wrong model — it is about a zero input. However beautiful an analysis built from an empty dataset, it is not knowledge, it is ornament. At 58, I have learned that tactics change, but denominators rarely lie. Per 90, per shot, per pass — as long as these denominators stay honest, the analysis stands. And the day the input goes blank, the model quietly begins to lie, and we never notice. So what is the signal for my readers in this transfer window? Simple. Whenever an analysis or a rumour reaches you, first ask — where did the input come from? Is the number verified, or a blank cell arranged in a handsome format? A club that decides by fee, wages, release clause and contract years beats the market's noise; a club that reads only headlines falls into the zero-input trap. Which clubs break and which thrive next season, I may well tell you in advance — but only when I hold complete, verified data. I do not make predictions that stand on blank cells. So the question turns to you: does your belief stand on data, or on a handsome table?