The Silent Spreadsheet: When a Football Analytics Pipeline Returns Nothing

One October morning in Liverpool, I reopened the report I had framed the night before. Nine sections, grey scaffolding fully built, tables aligned to the millimetre, a six-row risk matrix, a field for the data source, a field for the verification timestamp. The structure was so compliant that an automated system could have stamped it "processed" and moved it along. And every content cell was empty.

Title: none. Source: none. Information points: empty list. Entities involved: unidentified. Time sensitivity: not assessed. Source quality: unjudgeable, because the source field itself was blank.

What I held was a product flawless in form and hollow in substance. Had I written on regardless — filling the blanks with guesswork, naming a club never mentioned in the underlying data, drawing a chart with no provenance — I would have produced the most dangerous artefact in this profession: a lie that reads beautifully.

There was a time I stood over a spreadsheet and felt I was witnessing a miracle at Anfield. But that October morning taught me the opposite, colder lesson: data only whispers when it actually exists. When data vanishes, the only thing left speaking is silence.

A pipeline with nothing to pass on

To understand why an empty file deserves an article, you have to understand how football's data pipeline works. At the lowest layer sits event data: every pass, shot and duel recorded with coordinates and timestamps. Above that sits tracking data, capturing the movement of twenty-two players and the ball in fractions of a second. Then comes the modelling layer: derived metrics such as expected goals, passes allowed per defensive action, field tilt, accumulated chance value. Finally comes interpretation — where a human turns numbers into decisions: whom to buy, whom to sell, how to play on Saturday.

Such a pipeline can break at any joint. A source page sits behind a paywall. Content renders through JavaScript so the scraper retrieves only a hollow skeleton. A numbering error in the queue lets an empty document pass the validation gate because its structure is valid. The terrifying part is this: the system reports success. No red flag. No exception. Just a document that looks respectable and contains nothing.

Data engineers call this a silent failure. It differs entirely from a loud one. A crashing server is known instantly. A model returning an empty but well-formed output slips quietly through every gate, and by the time someone notices, it has already spawned three reports, one analysis and a personnel decision.

I asked myself: what if this had been a club's pipeline? What if that empty file were a scouting report on an eighteen-year-old in South America, and the only person reading it saw the conclusion pre-written in the template? The answer cost me several nights of sleep.

The tactical slice: when a model cannot see a corner kick

In 2026, I analysed all sixty-four matches of the World Cup in Russia using an expected-goals model I had built myself. My model predicted France as champions straight from the group stage, because their chance-creation index was the highest of the tournament, around 2.4 expected goals per match. France did win.

But Croatia reached the final, and I was exhausted. My article was mocked for claiming Croatia had a low chance-creation index and had advanced on luck. I hid in a library for two weeks, reopened the entire dataset, and found the flaw: my model ignored corners. Croatia that year scored a significant share of their goals from set pieces, and my model treated every corner as an ordinary shot in probability terms.

That lesson shaped how I have written for seven years since. xG is a revolution, but every revolution needs time to be accepted. More importantly, every revolution has a blind spot. The tidier the model, the harder the blind spot is to see.

Looking back at the past English season, tactical blind spots have shifted. If 2026 to 2026 belonged to high pressing, with the passes-allowed-per-defensive-action figure dropping below 9, then the recent period belongs to purposeful control. Liverpool under their Dutch manager won the 2026-25 Premier League with 84 points, playing with less chaos and fewer frantic transitions, but with markedly better positional control. The team still won. The way they won was entirely different.

For anyone working with data, that shift poses a hard question: if the old metric set scores both styles highly, what exactly is that metric set measuring? A metric that measures everything measures nothing.

I once calculated Liverpool's average pressing index in 2026-18 at 8.2 — the lowest in the league — while Manchester United in the same season sat at 15.7. That number said a great deal about Jürgen Klopp's philosophy. And the 4-3 win over Manchester City at Anfield in January 2026 was living proof: when pressing works properly, it stops being a tactic and becomes a mental state for the whole stadium.

At Anfield I learned that belief is also a variable. But belief cannot be measured with instruments. That is why I always write a "limitations of this analysis" section at the end of every piece, whether readers get to it or not.

The transfer market: a young-price bubble and the cost of fake certainty

Every figure in a transfer ledger is a life waiting to be written.

In January 2026, Chelsea paid around 106.8 million pounds for Enzo Fernández, a twenty-two-year-old Argentine midfielder with half a season of European football behind him at Benfica. In the same window, Mykhailo Mudryk moved from Shakhtar Donetsk to Chelsea for a reported 62 million pounds, potentially rising to 89 million with add-ons. In August 2026, Brighton sold Moisés Caicedo to Chelsea for 115 million pounds, then a record fee for a transfer between two Premier League clubs.

Three deals, three valuations, one shared pattern: clubs paying for unproven potential at the largest scale the game has ever seen.

For anyone sitting at the negotiation table, this is not hard to explain. Performance pressure means boards dare not wait. A twenty-two-year-old who has proven himself over thirty matches in a top European league becomes scarcer than gold. And when an asset is scarce, price is set not by quality but by the fear of someone else buying it first.

I call that the panic premium. It appears in no valuation model, yet it features in almost every major summer deal.

European football has also noticed the structural loophole. In December 2026, the Premier League capped contract length at five years for accounting purposes, after UEFA did the same from summer 2026. The move targeted the trick of stretching contracts to seven or eight years to spread a transfer fee across many seasons.

But closing an accounting loophole does not cool the market for young players. It pushes pressure into other structures: performance add-ons, sell-on clauses, buy-back rights. The spreadsheet still looks beautiful. Only the footnote gets smaller.

Results and the opinion cycle: from 46 per cent to 39 per cent

In March 2026, world football stopped. Liverpool were twenty-five points clear at the top and all but certain champions. The season was suspended. I lost faith in my own profession: if data could not predict a pandemic, what was data for?

I wrote three drafts and deleted them all.

When football returned in June 2026 in empty stadiums, one number emerged and haunted me for years: the home win rate in the Premier League fell from roughly 46 per cent to 39 per cent. Something invisible — the roar of twelve thousand people behind a goal — turned out to be worth seven percentage points of home advantage.

Empty stadiums did not distort the data, but they made the truth hollow. Every points model built on historical data became quietly biased, because it carried a variable that had vanished from the equation.

I spent weeks alone redefining my analytical framework. Since then, every metric I use must answer one question: under what conditions was it measured?

The current opinion cycle runs on the same logic. When a team wins three straight games while generating less expected goals than their opponents, the press calls it character. When they lose the next three with the same underlying numbers, the press calls it a crisis. One dataset, two stories, neither wrong in its arithmetic, and both missing the core question: is the sample large enough?

In a world of long seasons, the awakened can only rely on their own spreadsheet. But a spreadsheet is only honest when the person building it is honest with himself.

The league landscape

The Premier League today is a tiered system with an income gap wider than any other league in Europe. Six clubs leading on broadcast revenue, commercial income and squad value form a tier of their own. The rest compete along two routes: a smarter business model, or new ownership capital.

Brighton are the clearest example of the first route. They buy in undervalued markets such as Ecuador, Paraguay and Denmark, develop players in a data-rich environment, and sell at large multiples. Brentford follow a different version of the same philosophy, optimising every buy-and-sell decision at the lowest possible cost.

But there is a rarely stated rule: the buy-low-sell-high model only works when there is a buyer further up. When the top tier tightens spending, the whole chain below contracts within months. The most recent summer showed this plainly: several mid-tier clubs held players they could not sell at the price they had expected.

Deeper still, multi-club ownership networks are reshaping talent flow. A player can belong to a group in the Americas, be loaned in Belgium, developed in Austria, then sold to a Premier League club. For data analysts, this is a far more complex accounting and incentive problem than a simple transfer.

Rules and compliance: the grey zone inside a clause

The Premier League has profit and sustainability rules. Their enforcement over the past two seasons has produced notable precedents.

In November 2026, Everton were docked ten points for breaching the rules across the three-season cycle ending in 2026-22. In February 2026, that was cut to six on appeal. In 2026-24, the club received a further deduction, taking its total for the season to eight points. In March 2026, Nottingham Forest were docked four points for a similar breach, and their appeal was later dismissed.

On another front, in February 2026, Manchester City were referred on more than a hundred charges of breaching the league's financial rules. The proceedings have stretched across multiple phases and remain a long-term monitoring story.

These precedents matter for a technical reason: they turn a financial clause into a direct competitive variable. A club can lose its top-flight status because of a bookkeeping entry from three years earlier. For transfer analysts, this is the hardest variable to model, because probability depends on a legal process, not on form.

This is where my years of watching matches collide with my own limits. I can model a team's chance of taking points away from home. I cannot model a panel's chance of publishing a verdict on a given day.

The same holds for refereeing disputes. The space for subjective judgement inside video review is far larger than the public imagines. The phrase "clear and obvious error" is itself a vague clause. Clear to whom? Obvious to what degree? From a 60-degree angle, or from the view of someone sitting two metres from a monitor?

In September 2026, during Tottenham against Liverpool, a legitimate Liverpool goal was disallowed because of a communication failure among officials, and the refereeing body had to publicly admit the error. The released audio showed the problem lay not in technology but in people, in process, in one sentence misunderstood between two men who both wanted accuracy.

Semi-automated offside technology has been introduced to narrow such grey zones. Yet even if the offside line becomes an absolute number, the final call on a challenge inside the box still belongs to a human. Numbers can be precise. Judgement cannot.

Management and the dressing room

Football analytics contains a paradox rarely discussed: the denser the data, the deeper the shadow. We can measure distance covered, touches, line-breaking passes. We can barely measure a captain's authority in the dressing room.

Summer 2026 was a case study in large-scale succession. Jürgen Klopp left Liverpool after nearly nine years, leaving a structure shaped in both philosophy and personnel. The board brought in Arne Slot, and in parallel restructured the football hierarchy with the return of Michael Edwards in a senior executive role from March 2026.

This is the model I expect to spread: technical power detached from the manager and anchored in the system. When Klopp left, the style changed, but results did not collapse. Liverpool won the title in 2026-25. That is the most accurate measure of a durable organisation.

In the opposite direction, many clubs remain so dependent on one individual that every metric becomes noise when that person leaves. For data analysts, this is a lesson in age curves and contracts: a thirty-one-year-old with two years left is a completely different equation from the same player at twenty-eight with four years left. The spreadsheet does not lie. But the spreadsheet does not ask questions either.

The risk profile: the biggest risk is the one you cannot see

When I build a risk matrix for a club, I split it into six groups: sporting, financial, personnel, regulatory, reputational and systemic.

The last is the most undervalued. Systemic risk is when the data used for decisions is wrong, missing, or misread — and nobody knows.

A club can lose a season to a bad signing. A club can lose a decade to a broken model nobody questions.

Back to the empty report on my desk. Its content risk is zero, because it contains no content. Its process risk is maximal, because it can pass every control. The only thing preventing harm is a gate requiring at least one information point and a non-empty source field.

That gate, in my profession, matters more than any model.

Media and the expectation gap

Every transfer window I receive hundreds of rumours from readers. Most have no source, no journalist named, no credibility tier. For professionals, this becomes a source-ranking problem: a report from a journalist with direct agent access differs entirely from one recycled from another article.

The gap between market expectation and objective assessment is where I find the most distortion. A young player scoring five goals in seven games will be valued above his true worth. A young player scoreless in seven will be valued below it. Both are symptoms of the same disease: treating a small sample as large evidence.

The emotional cycle of English football media is shorter than a player's development cycle. That is why I write slowly. Not because I enjoy slowness, but because writing at the speed of public opinion would force me to say things I have not verified.

Those who are right ahead of their time always pay in solitude. In English football analytics, the first generation — the people who carried spreadsheets into boardrooms and were laughed at — paid that price. Today every club has an analytics department, and few remember that for a long stretch, the person raising expected goals in a transfer meeting was treated as an eccentric. xG is a revolution, but every revolution needs time to be accepted.

The industry transmission chain

At a macro level, football runs as a transmission chain with four links: academies and talent supply, clubs and competitions, broadcast and commercial rights, and finally derivative markets including betting, commercial data and digital content.

A change at the first link takes years to reach the last. A change at the last can travel back up within months. That is why prices for young players spiked after data platforms and digital media exploded: demand for content grew faster than the supply of talent.

The 2026 World Cup in the United States, Canada and Mexico, with forty-eight teams, is an example of expanding the middle link to serve the last. More matches, more markets, more broadcast hours. But football quality does not multiply with the number of teams. For data analysts, an expanded tournament creates a completely new sampling problem: weaker participants mean a more skewed score distribution, and every predictive model built on previous tournaments needs recalibration.

I have watched matches across many major tournaments, and every time a tournament expands, I rewrite the front end of my model. Not because football changes, but because the definition of a representative sample changes.

The contrarian angle: correlation is not causation

One mistake recurs across the analytical reports I read: turning correlation into causation, then causation into a rule, then using the rule to make a decision.

Teams that run more win more. True in some samples, false in many others. Teams that pass more control better. True until they meet a side that concedes the ball deliberately and counterattacks.

In the worst case, an analyst finds a data pattern supporting the conclusion they wanted, then presents it as an objective discovery. I have done that, and I paid with two weeks in a library.

Return to the empty report. It is an exact metaphor for a larger problem: we are building systems that produce conclusions faster than we can verify them. Correct structure, correct format, correct process. Nobody checks the content.

With VAR, this happens at national scale. A tool designed to reduce error has created a new category of error: error in the review process. The phrase "clear and obvious error" appears to limit intervention, but in practice opens a grey zone filled by individual judgement. Nobody is wrong. Yet outcomes are inconsistent.

That is what I want readers to carry away: a system can be structurally correct and substantively meaningless. A clause can be verbally sound and operationally vague. A spreadsheet can be beautifully formatted and empty of truth.

Signals to watch in the next round

As this cycle reaches its decisive phase — European domestic leagues sprinting to the finish, relegation races settled by both points and financial filings, and the 2026 World Cup approaching — there are four signals I will track in my spreadsheet.

First, the ratio between points won and expected goals among the leading six. If the gap keeps widening, finishing quality is creating an advantage harder to close than any tactical system.

Second, the adaptation speed of clubs taken over by new managers in their first six months. This is the best test for distinguishing a club with a system from a club with one talented individual.

Third, compliance behaviour in major transfers. The five-year contract cap has turned every huge fee into immediate accounting pressure. The market's answer will lie in add-on structures, not headline totals.

Fourth, the source-verification quality of the analytical systems increasingly appearing across football. A good model is one that knows what it does not know.

I still end every report with a single line of self-interrogation: if my writing were a data pipeline, where would its final gate sit?

For me, that gate sits in my own silence. When there is no data, the most honest thing I can do is close the file and tell the reader there is nothing yet to say. Data whispers, and those who listen will hear the miracle. But when the spreadsheet falls silent, the true listener is the one who bows their head and writes in the log: this space is still empty.