The youth-fee bubble and the scouting reports with no data in them

On 31 January 2026, Chelsea completed the signing of Enzo Fernández from Benfica for €121 million. Enzo was 22, had played fewer than twenty matches in European football, plus seven World Cup games in Qatar. The scouting report I read the next morning ran to four pages, was printed in colour, carried three charts, and closed with a single line: "unlimited development potential". None of the charts stated its denominator. No line mentioned how small that denominator was.

I sat at my desk in Liverpool, reopened my spreadsheet, and found something familiar: the transfer market runs on reports that are formally complete and substantively hollow. Literally hollow. A ten-section analytical framework, every section titled, every cell carrying the phrase "insufficient information to conclude". People still sign hundred-million-euro contracts on documents like that, and nobody finds it strange.

I entered the trade in 2026, in the sports department of Belgrade Television. There was no xG then, no PPDA, no live data feed. We had a notebook, a pen and one rule: record what the eye saw, not what you wanted it to see. Thirty-five years later that rule is still my entire methodology; only the pen has become a spreadsheet.

In 2026, working as a transfer market administrator in Liverpool, I calculated the team's average PPDA under Juergen Klopp at 8.2 — the lowest in the Premier League — against 15.7 for Jose Mourinho's Manchester United. I wrote a long piece on gegenpressing and was told it was "too mechanical". Liverpool's 4-3 win over Manchester City on 14 January 2026 was the first time I saw the Anfield crowd operate exactly as a data table had predicted. I do not retell that to praise myself. I retell it because it taught me the opposite lesson: the denominator is the frightening part, not the conclusion.

Every average has two halves. The first is the value. The second is the number of observations. Football analysis spent twenty years arguing about the first half and said almost nothing about the second. A midfielder averaging 2.4 key passes over thirteen matches can fall to 1.1 over thirty, and that does not mean he has become worse. It means the denominator changed.

The Enzo Fernández case is a clean test. Before joining Chelsea he had played 13 Primeira Liga matches for Benfica, 6 Champions League group games and 7 World Cup matches. Twenty-six fully recorded appearances in total, only 19 of them on European soil. Divide €121 million by those 19 matches and you get roughly €6.4 million per game. That division is crude, and I know it is crude. But it forces the eye onto the denominator instead of the final figure in the newspaper.

Mykhailo Mudryk is the second case. On 15 January 2026 Chelsea signed him from Shakhtar Donetsk for €70 million plus up to €30 million in add-ons. Mudryk was 22, with fewer than thirty Ukrainian Premier League appearances and six Champions League group games. Antony joined Manchester United in September 2026 for €95 million after two full Eredivisie seasons and a notable Champions League campaign. Each deal carried a different denominator, and none of the three passed one hundred top-level matches.

Set Jude Bellingham beside them. In June 2026 Real Madrid paid €103 million plus add-ons for a 19-year-old. But when he signed, Bellingham had 44 appearances for Birmingham City and 132 for Borussia Dortmund — more than 170 club matches, plus two Champions League seasons and a World Cup. Similar fee. A denominator almost ten times larger.

Joško Gvardiol runs the same way. Manchester City paid €90 million for him in August 2026, by which point he had two seasons at RB Leipzig, a European campaign with Dinamo Zagreb and a full World Cup as the left-sided centre-back in a back three. High fee. Deep evidence base.

Accounting makes the bubble more durable than it looks. A €121 million contract spread across eight and a half years is roughly €14 million of amortisation a year, plus wages. Under the Premier League's profitability and sustainability rules that amortisation registers as a legitimate cost, which gives clubs an incentive to stretch contract length and thin the annual charge on the books. The balance sheet looks healthier. The risk does not vanish; it is sliced smaller and pushed forward.

I keep a private sheet I call "top-level minutes". For every player under 23 valued above €50 million, I log the minutes played in a top-five European league or in the knockout rounds of a European competition, up to the day of signing. I have tracked it since 2026. The trend over the last four years is uncomfortable: transfer values have grown about three times faster than verified top-level minutes.

The transfer market is pricing broadcast minutes, not verified top-level football minutes.

Enzo's seven matches at the 2026 World Cup were watched by billions. His thirteen Primeira Liga matches were watched by tens of thousands. Same player, same skill set, two levels of exposure — and only one of them entered that four-page report. The auction mechanism does the rest: when three clubs see the same moment replayed in slow motion three times, the price stops tracking productive capacity and starts tracking the fear of missing out.

In March 2026, when football stopped for the pandemic, I lost faith in my own method. Liverpool were 25 points clear of Manchester City and the season was suspended. When football returned in empty stadiums, I measured home win rates falling from 46% to 39%. Same squad, same tactics, one environmental variable removed, and the results shifted. Since then every analysis of mine carries a fixed section: the conditions of measurement.

This is where I have to argue against myself. A small denominator means a very wide confidence interval around any judgement about a player, and a wide interval does not mean the judgement is wrong. Some players need nineteen matches to reveal a ten-year career. The problem is that the interval never gets written into the report, and the report then gets used as evidence for an investment.

Data whispers; those who know how to listen will hear the miracle.

But first someone has to be willing to listen.

There is a misreading I meet constantly: "expensive young players tend to fail". It sounds reasonable and it is statistically wrong, because the group called "expensive young players" is really two different groups bundled together. The first has a deep evidence base and is priced highly because of that evidence. The second has a thin evidence base and is priced highly because of one moment. The correlation between price and failure does appear, but the cause sits neither in the price nor in the age. It sits in the denominator.

I fell into that trap myself, by a different route. In 2026 I built a homemade xG model to analyse all 64 World Cup matches in Russia and concluded that Croatia were going far on luck. Croatia reached the final. I spent two weeks in a Liverpool library, rewatched the data, and found the fault: my model ignored set-piece sequences from corners. One omitted variable can produce a very confident conclusion. xG is a revolution, but every revolution needs time before people accept it — including the people who built it.

The same linguistic disease shows up elsewhere in football: the VAR room. "Clear and obvious error" is a phrase that sounds like a criterion, yet in operation it hands the decision back to the on-field referee's feel. The line between "clear enough" and "not clear enough" is not measured in millimetres but in something nobody can audit. A system built to reduce subjectivity generated a new layer of subjectivity, and that layer is harder to argue with because it is wrapped in technical language.

I learned at Anfield that belief is a variable too.

The person who is right before his time always pays in solitude. In a world of long seasons, the one who wakes up early can only rely on his own spreadsheet.

The signal I will watch in the next transfer window sits in the ratio between verified top-level minutes and age at the point of signing, not in the total fee. Clubs that build the habit of recording denominators will buy cheaper than the rest, and they will not tell anyone why. Meanwhile the ten-section reports with "insufficient information" in every cell will keep being printed, keep being signed, keep being filed in a folder nobody reopens.

What I want to know: when did you last look at a transfer data table and ask what its denominator was?