The Annual Season: Conclusions Written on an Empty Data Set

At the 88th minute, the stats board lit up on the big screen at the eastern corner of the stand. A neat row of numbers: xG 2.31 – 0.47, followed by a small line about passes into the box. The man beside me, beanie pulled down over his ears, turned and said one short sentence: "They deserved to lose." I looked back at my notebook. Fourteen lines of writing, not one of them a number. The third line read: "goalkeeper holds the ball, stands still for four seconds, eyes on the B stand." The ninth line: "left-back puts his hands on his knees in the 71st minute, nobody tells him anything." The eleventh line: "a child in the front row waves, nobody waves back."

I did not write down the goals. I wrote down the silence before the ball hit the net, the way the whole stand breathed in on one rhythm and then forgot to breathe out.

That night I went home, opened the data platform, and saw what anyone who has been in this trade long enough has already seen: the sample was three matches. Two hundred and seventy minutes. And on the simulated league table, a zone had already been shaded red and named "the relegation area."

Three matches. The annual season had barely opened its notebook, and someone had already written the final chapter.

The candle lanterns that year did not light up Anfield, but they lit up an entire season without spectators. I repeat that line because it is the foundational lesson for anyone who reads football for a living: what we see on the pitch has never been the whole story. In June 2026, Liverpool won the Premier League for the first time in thirty years, and the city had no trophy parade. I drove around Liverpool on the night of 25 June, watching people stand at their front doors, lighting lanterns, singing "You'll Never Walk Alone" through phone speakers. There was no match to commentate on. But there was a season to read.

That is why I always begin with a breath, not a spreadsheet.

Context: a season with no climax, only pressure

The annual season differs from a cup competition in that it has no single point of gravity. There is no final for everyone to hold their breath for. There are thirty-eight rounds, and across those thirty-eight rounds, every team must play at an intensity the human body was not designed to sustain for nine consecutive months.

In England, a club playing in the Premier League, the League Cup, the FA Cup and European competition can reach March having already played forty-five matches. Add international windows, and the number passes fifty. That is why analytics departments in England split a season into five phases: warm-up, accumulation, midwinter crisis, regeneration, and the final sprint. Each phase has its own set of metrics, and those sets cannot be used interchangeably.

The first three rounds are the warm-up phase. During this phase, the PPDA metric — passes allowed per defensive action — typically runs fifteen to twenty percent above its full-season average. The reason is simple: teams have not yet found their pressing rhythm, the lines have not yet locked their spacing, and many players are still at the physical level of the third week of pre-season.

What does that mean? It means every tactical conclusion drawn from the first three rounds is describing a team that does not yet exist. This is what I call an empty data foundation: a set of figures that looks complete but carries no information sufficient for a conclusion.

Based on my experience watching matches across many seasons, I can state something that is uncomfortable for the data-media business: most xG tables published after round three are meaningless as forecasts. They only acquire meaning once the sample passes ten matches. Yet they are still published, still cited, still used as the basis for commentary pieces with very confident headlines.

There is one professional detail I want to share. In European analytics departments, there is a type of report nobody wants to sign: the blank report. It is the pre-match briefing that comes back from the data department with exactly one line of conclusion — insufficient information to assess. Professionally, it is the most honest report in the building. Politically, it is the report that gets the person who signed it asked some very uncomfortable questions.

And in the annual season, when there is a match every week, the pressure to fill the page is always greater than the pressure to be right.

The core: three markets that are mispricing

From that empty foundation, I want to move into three specific markets of modern football where price and actual capability have drifted far enough apart to be measurable.

First, goalkeepers and the myth of the feet.

In 2026, Chelsea paid 71.6 million euros for Kepa Arrizabalaga, at the time the highest fee ever paid for a goalkeeper. The stated reason was not shot-stopping, but the ability to participate in build-up play from the back. That same year, Liverpool paid 62.5 million euros for Alisson Becker, and Manchester City had already paid 40 million euros for Ederson in 2026. All three deals sat inside the same current: the goalkeeper had become a tactical position rather than a purely defensive one.

I do not deny the value of distribution. I object to the share of the valuation it now receives.

Look at the structure of the two metric families. Shot-stopping — measured by save percentage or by post-shot expected goals — is a highly stable metric across seasons, weakly dependent on tactical system, and repeatable. Distribution — measured by progressive passes, long-ball accuracy, involvement in build-up sequences — depends enormously on three variables entirely outside the goalkeeper's control: how the opponent presses, the quality of his own centre-backs, and the tactical instruction of the coaching staff.

Put differently: a goalkeeper with beautiful distribution numbers inside Manchester City's system can post very poor distribution numbers after moving to a club that faces a high press every week. But his shot-stopping travels with him for his entire career.

What the transfer market is doing is paying for a skill that depends on the system, while discounting a skill that depends on the goalkeeper himself.

I have sat through goalkeeper training sessions at two different clubs. The structure was nearly identical: about twenty minutes on distribution and combination play with the back line, and more than an hour on reflexes, close-range handling, aerial work and one-on-one reactions. The time split in training does not match the valuation split in the market. That is a paradox I have never seen explained satisfactorily.

Second, the young-player price bubble.

In January 2026, Chelsea completed the signing of Enzo Fernández for 106.8 million pounds, making him the club's most expensive player at the time. Enzo had then played fewer than thirty matches in European football for Benfica. In the same window, Mykhailo Mudryk arrived for 70 million euros plus 30 million euros in add-ons, after roughly forty matches at the highest level for Shakhtar Donetsk.

By August 2026, Chelsea broke the British transfer record again, paying 115 million pounds for Moisés Caicedo.

The transfer market does not sell players; it sells dreams priced by fear. Fear of missing a player who will shine somewhere else. Fear of falling behind in a race nobody knows the finish line of. Fear of being criticised for having done nothing.

There is an unwritten rule that some European analysts have proposed and that has never been adopted: do not pay a nine-figure fee for a player who has not yet played fifty matches at the top level. The rule sounds arbitrary, but it comes from a very specific logic. Below the fifty-match threshold, the sample is too small to distinguish between a genuinely elite player and a player at the peak of a short form cycle. Above it, forecast probability rises sharply.

The problem with a nine-figure contract is not the number itself. It is the chain reaction. A 115 million pound fee amortised over eight years equals roughly fourteen million pounds a year in the accounts. But the accompanying wage does not amortise — it immediately becomes the reference wage for the entire dressing room. Within six months, three key players will ask to renegotiate. Within twelve months, the club will have to sell someone important to balance the books.

That is why I call this a naked gamble: the club is not betting on the player, it is betting on the endurance of its wage structure over the next four years.

Vietnamese football is also touching the edge of this problem, at a much smaller scale. When a V.League club sells a twenty-year-old to Japan or Korea, the fee is usually somewhere between a few hundred thousand and a few million US dollars. But that fee as a percentage of the club's total budget is far higher than a nine-figure deal is for a European club. The systemic risk here is larger, even though the absolute numbers are smaller.

Third, esports and a career span stolen away.

Here the story is entirely different. If a footballer can play at the top from seventeen to thirty-five — roughly eighteen years — an esports professional typically gets three to five years at the peak.

Lê Quang Duy, known by his in-game name SofM, is one of the exceptional cases in Vietnamese esports, sustaining an international playing career for nearly a decade in League of Legends. Đỗ Duy Khánh, competing as Levi, is a similar case in the jungle role. But they are the exception, not the rule.

Esports is where young men without divine feet still touch glory with their fingertips. And precisely for that reason, that glory arrives earlier and vanishes faster than in any other sport.

A professional player in Vietnam joins the first team at eighteen. Peak performance falls between twenty-one and twenty-three. Reflexes begin to decline from twenty-five. And by twenty-seven, most have lost their starting spot.

In that window, what do they have? Season-by-season short contracts, wages that fluctuate with results, no mandatory pension fund, no players' union, no career-transition programme. A footballer retiring at thirty-five usually has eighteen years of accumulation and an industry network thick enough to move into coaching, scouting or broadcasting. An esports professional retiring at twenty-five usually has seven years of accumulation and an online following that can evaporate within six months if content is not maintained.

That is why I say that esports talent development is running at least ten years ahead of its post-retirement support system.

The contrarian angle: the blind spot of collective memory

Now I want to return to those fourteen lines in my notebook.

Our collective memory of a match is built from moments that have images. A shot. A save. A shattered microphone in the technical area. Things without images are not retained: the four seconds a goalkeeper stood still, a left-back with his hands on his knees in the 71st minute, a child waving with nobody waving back.

The blind spot sits here: we built an entire modern analytics apparatus to measure what can be measured, and then quietly accepted that what is measured is the whole match.

In the annual season, that blind spot is amplified exponentially. There is a new match every week, a new dataset every week, and every week a fresh batch of conclusions drawn from a sample that is still not large enough. After thirty-eight rounds, we have thirty-eight sets of conclusions, most of which contradict each other.

And the striking thing is that nobody is held accountable for the conclusions that were contradicted. There is no mechanism recording that in round four, an expert declared a club certain to be relegated.

Football is the only place where adults are allowed to cry like children without needing to explain. Perhaps that is also why football is the place where people are allowed to conclude without needing evidence.

I no longer chase the ball the way Mbappé does, but I have learned to chase its story. And the true story of an annual season is not in the table after round three. It is in the things that only appear after round twenty: the tactical current of a team that has found its structure, the physical pressure accumulated month by month, and the refereeing controversies that only mean something when placed side by side in a long sequence.

There is a recent event in Southeast Asia I want to bring back because it illustrates this point sharply. On 5 January 2026, the Vietnam national team walked into the second leg of the ASEAN Cup final at Rajamangala Stadium in Bangkok. Vietnam lost that second leg 2-3, but won the title 5-3 on aggregate across the two legs. In that match, Nguyễn Xuân Son scored twice and then suffered a broken leg.

Read only the second leg's stats and you see a losing team. Read the whole tie and you see a champion. Read only the tournament's statistics and you see a collection of numbers. Sit in Bangkok that night and watch how the Vietnamese players endured the final fifteen minutes, and you see something else entirely — something no dataset can transmit.

What I mean is not that data is useless. What I mean is that data only has value when the person reading it knows what is missing.

Takeaway

The annual season will keep moving. There will be thirty-five more rounds, thirty-five more weeks of numbers, thirty-five more times we are forced to choose between a fast conclusion and an honest answer.

In this trade, the hardest sentence remains the one every analytics department dreads writing: insufficient information to assess.

Every season that passes is a book being closed; the careful reader will find himself inside it. I hope that when this season closes, there will be a few more careful readers than before.