Cristiano Ronaldo cost €94M in 2009.

Hugo Ekitiké cost €94M in 2025.

The conclusion usually follows immediately:

Inflation.

And that is true. Just not in the way most football fans think.

Something clearly happened. Because “inflation” is doing an enormous amount of work in that argument.

Behind that one word, however, sit three related but very different concepts — and football debates quietly mix them together.

1. The euro itself loses value. A euro today buys less than a euro in 2009. Ronaldo’s fee looks smaller every year without football changing at all — the measuring stick shrank.

2. Football’s own prices rise faster than everything else. The way housing or tuition can outrun general prices, transfer fees can outrun bread and rent. That is a property of the transfer market, not of the euro.

3. The player behind the fee changes. €95M bought a 24-year-old reigning world’s best in 2009 and a promising winger in 2022. If the product changes, the price comparison means something completely different — statisticians call this the quality-adjustment problem, and it haunts every inflation number you have ever read.

Three concepts.

One word.

So for today’s piece, I went through 28,000 paid transfers to separate the three and show what each one actually tells us about the transfer market. Then, at the end, I put them back together and return to the famous Ronaldo comparison.

Because the meme is not completely wrong. It is just much more misleading than it first appears.

Let’s get into it.

1. The money: your memory of prices is broken

Start with the boring one. The euro itself.

The cumulative euro-area inflation since 2009 is roughly 43%. That means any fee from 2009 is understated by almost half when you read it with 2026 eyes.

Here is what the famous world-record transfers look like once you restate every fee in today’s euros:

Look at the first five numbers. €107M. €131M. €135M. €133M. €138M.

For sixteen years, the real price of the most expensive player on Earth barely moved. The nominal record kept climbing. The real record really didn’t.

Then Neymar happened — €222M in 2017, about €287M in today’s money — and nothing has come close since.

Which means the real world record has been quietly falling for nine years, eaten by ordinary inflation. And the same correction reshuffles the top 20 all-time list:

Ronaldo’s 2009 move climbs from 24th back into the top 15. Zidane jumps from 46th to 17th — above Wirtz, above Havertz. And a group of recent €95M signings falls out of the top 20 entirely, because their euros are the cheapest euros on the list.

The first confusion

Comparing prices from different years without converting the money is not an opinion about football. Sure it will get you clicks on social media. But above all it is a measurement error.

And at the very top of the transfer market, fixing that one error explains more of the “fees went crazy” feeling than almost anything else.

2. The market: football’s own inflation

Now the more interesting question. After removing ordinary inflation, did football itself get more expensive?

This is the second concept — a relative price change — and it is where measurement gets genuinely tricky.

To measure it properly, borrow the tool economists built for exactly this job.

The CPI — the Consumer Price Index. It is the number governments actually publish when they announce “inflation was 2% this year”. And it works in a deliberately boring way. Statisticians define a fixed basket of goods — bread, rent, haircuts, train tickets — and then price the same basket, year after year.

Not the average thing people happened to buy. The same basket.

That fixed basket is what makes inflation numbers comparable over time: if the basket changed every year, you would never know whether prices moved or the shopping list did.

Football’s version is not a fixed basket. It is a fixed market rank.

Every season has a 10th-biggest signing, a 25th-biggest, a 100th-biggest. Those positions exist every year, no matter how many small deals enter the database. A crude football CPI.

In constant 2026 euros:

• the season’s 10th-biggest signing: €47M in 2014 → €71M in 2025 — +51% real

• the 25th-biggest: €27M → €54M — +103% real

• the 100th-biggest: €11M → €26M — +128% real

So football inflation is very real. It just lives in a different place than people look.

Here’s another way to illustrate the same idea. The figure below presents the number of transfers between €50M-€100M and above €100M.

In 2014 there were 8 transfers above €50M in today’s money. Recent seasons run at 21 to 29.

And in most seasons more than half of the buyers are Premier League clubs. Which raises the obvious question.

Where is all this money coming from?

Follow the money: the TV cheque

Mostly from one place. Television. And especially from England.

And since this whole piece runs on constant 2026 money, let’s hold the TV deals to the same standard and strip ordinary UK inflation out of them too.

When the Premier League launched in 1992, its broadcast rights were worth about £0.1 billion a season in today’s pounds.

The current cycle is worth about £3.8 billion a season. Thirty-eight times more. After inflation.

Three details in that chart matter for transfer fees.

First, the timing. Between the 2013–16 cycle and the 2022–25 cycle, the Premier League’s real TV income rose 52%. The real price of a top-tier signing rose 51% over the same window.

Coincidence? Same period, same magnitude — the fees track the cheque.

Second, the source. Since the 2022–25 cycle, international rights are worth more than domestic ones. The rest of the world is now funding the Premier League’s shopping.

Third — and easy to miss — the plateau. In real terms, the TV cheque has been roughly flat since 2019. The engine that inflated the transfer market has stopped accelerating. If fee inflation cools over the next few cycles, this chart is why.

Now translate the cheque into spending power.

Split it across twenty clubs and every Premier League club starts the season with roughly £150–175M of central broadcast income. Before selling a single ticket or shirt.

For scale: the median Premier League club’s entire wage bill is about €84M. The median club in Spain, Italy or Germany runs on €33–36M.

So the Premier League’s TV cheque alone hands each of its clubs more than the full wage bill of a typical Serie A or La Liga club — several times over.

And that explains where football transfer inflation comes from. The €25–70M tier is the shelf where Premier League clubs shop. Not just the big six — all twenty. Brentford and Bournemouth can pay fees that only champions could pay elsewhere.

Twenty clubs, armed with broadcaster money, bidding mostly against each other for the same pool of players.

Prices follow.

The second confusion

Sector inflation is about the market, and it cannot be measured with a naive average. Price fixed positions instead — the way a CPI prices a fixed basket — and the answer is clear. The 25th signing of the season costs twice as much in real terms as it did in 2014. That is football’s real inflation rate. And its engine has a name: the Premier League TV deal.

3. The product: the same money buys a different player

Now the third concept. The one almost nobody names, even though professional statisticians fight it every day.

When the price of “a phone” looks stable for a decade, is that zero inflation? The 2026 phone is not the 2016 phone. The product changed under the price.

Statisticians call this quality adjustment, and it is the hardest part of their job.

Football has exactly this problem. Because the thing a €100M fee buys today is not the thing it bought ten years ago. Here are three tests to illustrate the point.

Test one: rank the players at the moment of purchase

For every €20M+ transfer into a top-five league since 2017/18, I took the player’s season immediately before the move and ranked him against every top-five-league player in his positional family — a composite score built from position-specific output metrics.

The individual scores are not the point. The distribution is.

For the same inflation-adjusted fee, are clubs buying higher-ranked or lower-ranked players than before?

Lower.

In 2018–19, the median €60M+ signing ranked around the 88th percentile of his position. Today, closer to the 77th.

The share of genuinely top-decile players among those buys fell from 43% to about one in three.

And note where the slide happens: the €35–60M tier barely moved. The change is concentrated in the biggest cheques.

One caveat. This chart can only understate the shift due to a limitation in my data which only has Top 5 (England, Spain, Germany, Italy, France) leagues data. Because one big signing in five has no top-five-league season to rank at all.

João Félix. Enzo Fernández. Antony. Darwin Núñez. Frenkie de Jong.

Mostly from the same pipeline: Benfica, Ajax, Sporting, Porto.

Elite prices, paid before an elite body of evidence exists.

Test two: age

The market as a whole did not get younger — about half of all paid transfers involve a player 23 or under, stable for twelve years.

What changed is who the big cheques go to.

Until 2017, a €30M+ signing was less likely to be 23 or yonger than an ordinary transfer. The big fee was a prize for the established player. Youth was what you bought cheap.

The lines cross in 2018–19.

They have not crossed back (if it wasn’t for the 2024/25 season which kind of ruined my beautiful conclusion).

Test three: forget my models — use the Ballon d’Or

For every €60M+ transfer since 2014, one simple question.

Had the player already finished in a Ballon d’Or top 10 before the move?

Between 2014 and 2019, buying a certified superstar was normal. Suárez to Barcelona. Neymar to PSG. Ronaldo to Juventus. Hazard and Griezmann in 2019.

Between 2020 and 2022?

Zero.

Not one €60M+ transfer involved a player who had ever been ranked among the world’s ten best.

And the few that appear since are farewell tours: Kane at 30, Neymar to Saudi Arabia at 31.

Where did the superstars go? The Messis, Benzemas, Ramoses.

All of them eventually changed clubs. None of them appears in eleven years of paid-transfer data.

They left on free transfers — running down contracts and capturing the fee themselves, as wages and signing bonuses.

The third confusion

Quality change is about the product. A price that looks stable — or crazy — means nothing until you check whether the thing being bought is still the same thing.

In football it is not.

The €100M+ fee used to buy proof. Now it buys probability.

4. All three at once: the most famous price in football

Now we can run the full decomposition on the number everyone argues about.

Ronaldo, 2009, €94M.

Concept 1 — the money. €94M of 2009 is €135M of 2026. That part has nothing to do with football.

Concept 2 — the market. Add football’s own inflation: top-tier signings rose another +51% real just since 2014 (my fee data starts there, so whatever happened between 2009 and 2014 comes on top). A player of that caliber should cost €200M+ in today’s market. Not a precise valuation — a magnitude.

Concept 3 — the product. Players worth that kind of fee still exist. Transfermarkt currently values Yamal and Haaland at €220M, Mbappé at €200M — almost exactly where the arithmetic lands. But none of them is for sale. What actually tops each window is €120–140M real, spent on the 20-to-22-year-old who might become them.

Boom — that was football inflation 101.

The next time someone tells you prices went crazy — for players, for houses, for engineers, for eggs — ask three questions:

Did the money change? Did the market change? Did the product change?

Most football posts fail at least two if not all three checks in a single meme.

Now you know better.

Talk soon,

Martin

P.S. This post was AI-assisted for editing and code debugging. Thanks to AI, my productivity curve also looks like the Premier League’s TV deal. My revenue curve, sadly, does not.

His contract with Manchester City was running down, and the club’s opening offer would have left him earning less than before.

The person who would normally fight that battle for him — his agent — was in prison.

So De Bruyne did something unusual. He negotiated without an agent and hired a data consultancy to help build his case.

The consultancy was Analytics FC, then a six-person firm. The engagement was so unusual that the service effectively did not exist before De Bruyne asked for it.

The result came in April 2021: a new Manchester City contract reportedly worth around €80 million, representing a raise of roughly 30%.

The story itself is now well known. It’s even taught at Harvard Business School (HBS). What's less public and what interested me was something else:

What did the data scientists actually build?

The HBS case describes a series of analyses designed to answer four questions:

1. How good was De Bruyne relative to the market?

2. What would Manchester City have to spend to replace him?

3. Which other clubs could realistically afford and use him?

4. Most importantly: what was De Bruyne worth specifically to Manchester City?

So I tried to rebuild them all.

Everything below uses information that was available around the time of the negotiation. Performance data comes from Opta, wage estimates from Capology, and the analytical framework follows the methods described in the HBS case study.

The result is about as close as I can get, from outside the club, to reconstructing De Bruyne’s negotiation pack.

Let’s get into it.

Why De Bruyne needed data scientists in the first place?

Before rebuilding the tools, it helps to understand the negotiation process.

A player under contract is an asset. Manchester City had bought De Bruyne from Wolfsburg in 2015 for around €76M and signed him to a six-year contract.

But contracts are wasting assets. As expiry approaches, the transfer fee another club would need to pay falls. Eventually the contract reaches zero and the player can leave for free.

That clock creates leverage on both sides.

• The player can say: extend me now, or risk losing me for nothing.

• The club can say: you are getting older, so perhaps we let the contract run down and replace you.

That is why renewals often happen with roughly two years remaining, while both threats still matter.

The problem for De Bruyne was that most of the other advantages belonged to City.

The club had proprietary performance data. It had professional negotiators. Sporting directors handle contract discussions constantly; a player may do it only a handful of times during his career.

Normally an agent absorbs that asymmetry. De Bruyne did not have one.

There was another weakness in his position: he wanted to stay. By every account, he wanted to remain at City and win the Champions League there.

That is not an especially convincing walk-away threat.

So the negotiation needed something else.

Rather than argue simply that De Bruyne was one of the best players in the world, his side needed to answer a more useful question:

What was keeping Kevin De Bruyne worth to Manchester City?

That is where the data scientists come in.

Tool #1: Benchmark him against the market

Was De Bruyne already being paid like the player he was?

The obvious place to start is with the market. Take every attacking player in Europe’s top five leagues and compare two things:

• what they produce;

• what they are paid.

For output, I use expected goal contribution per 90 minutes: non-penalty expected goals plus expected assisted goals.

Why expected numbers?

Because a contract prices the future. Actual goals can swing because of finishing. Actual assists depend partly on whether teammates convert the chances you create. Expected metrics strip out some of that noise and get closer to the underlying quality of the chances a player generates.

I also use three seasons rather than one. A long-term contract should price a sustained level rather than a single unusually good or bad year.

Over 2018/19 to 2020/21, De Bruyne produced approximately 0.75 expected goal contributions per 90 minutes. That put him around the 98th percentile among attacking players in Europe’s top five leagues.

Now look at wages. The median attacking player in this pool earned around €1.9M a year. Roughly two-thirds earned less than €3M.

But wages do not rise smoothly with performance. For most of the market they climb slowly. Then, among the final handful of players (the Messis, Ronaldos, Neymars, Mbappés), they go almost vertical.

The top ten players — barely one percent of the sample — captured more than 12% of all wages paid to the 872 players in the pool.

De Bruyne sat right at that edge.

His estimated pre-renewal salary of around £350k per week — roughly €22M annually — placed him in approximately the 98.5th wage percentile.

So the first piece of the negotiation case is almost perfectly symmetrical:

• 98th-percentile output.

• 98.5th-percentile pay.

He was being paid like one of the best attacking players in Europe because he was producing like one.

And that conclusion survives more sophisticated measurement. Using the twelve-metric creative profile from my previous article puts De Bruyne in the 99th percentile. Using the full 43-metric PCA fingerprint puts him around the 96th. One metric, twelve metrics or forty-three: the exact number moves, but the conclusion does not. De Bruyne lived at the extreme right-hand edge of the market.

Where normal salary benchmarking breaks

De Bruyne was already being paid like an elite player. His wage rank was almost perfectly aligned with his performance rank.

But there was still another level above him.

Messi, Ronaldo and Neymar were not simply top-percentile players. They occupied a different part of the wage distribution altogether — the extreme superstar tier where salaries stop rising gradually and jump sharply.

That was effectively the territory De Bruyne was trying to move toward.

So the question was no longer whether he deserved to be among Europe’s best-paid players. The data already showed that he was.

The harder question was: how much more should a club pay for the difference between an elite player and a true superstar?

To see what a conventional salary-benchmarking approach would say, I fitted a basic wage model using player-seasons from 2017/18 through 2020/21. The model learns what the market typically pays based on:

• output;

• minutes played;

• age;

• league;

• the employing club’s wage bill.

Economists would call this a hedonic model. HR departments use essentially the same logic when benchmarking salaries against comparable jobs.

The model explains around 70% of wage variation across Europe.

Then I ask it to price De Bruyne. Its answer: about €11M per year. His estimated actual salary was already more than twice that. By the model’s logic, De Bruyne was massively overpaid.

Change the specification and the problem remains. Use sustained three-season output: around €12M. Use the PCA-based profile score: around €14M. Still nowhere close.

Then look at the other players the model calls overpaid: Messi. Neymar. Ronaldo. Griezmann.

That tells us something important. When a model systematically concludes that the most valuable superstars in European football are all being paid far too much, it probably has not discovered a giant market inefficiency.

It is missing the way superstar markets work.

The value of moving from the 56th percentile to the 60th percentile is not the same as moving from the 95th to the 99th. At the top of football, small differences in quality can decide league titles and Champions League runs.

The prize money, commercial value and scarcity attached to those differences are enormous. A regression fitted across the whole labour market smooths that away.

And that is why ordinary salary benchmarking was never going to justify De Bruyne moving into the very top wage tier. To make that case, the negotiation had to stop asking what comparable players earned and start asking a different question:

What would losing Kevin De Bruyne actually cost Manchester City?

That leads to the second tool.

Tool #2: Price the replacement

What would Manchester City have to spend if De Bruyne left?

If City refused to pay De Bruyne, eventually they would need someone else to perform his role.

So who could actually replace him?

I started with every midfielder and forward in the top five leagues who played at least 15 full-match equivalents in 2020/21 — around 700 players. For each player I calculated twelve creative metrics per 90: key passes; expected assists; expected assisted goals; passes into the penalty area; passes into the final third; progressive passes; shot-creating actions; goal-creating actions; progressive carries; penalty-box touches; progressive passes received; non-penalty expected goals.

These metrics live on different scales, so each one is converted into a z-score: the number of standard deviations a player sits above or below the population average.

That gives every player a twelve-number creative fingerprint. Then I measured the distance between each fingerprint and De Bruyne’s. The smaller the distance, the more similar the player. And this is where the negotiation gets interesting.

The closest genuine substitutes fell into two groups.

The first group contained players such as Messi and Neymar. They were clearly good enough. They were also already on superstar salaries or carried astronomical acquisition costs.

The second group contained much cheaper statistical lookalikes. Players such as Ruslan Malinovskyi could resemble De Bruyne in the shape of their output while earning a tiny fraction of his salary. But shape is not level. They were producing less, at weaker teams, without evidence that the performance would survive the jump to a title-winning environment.

The market understood that difference.

And perhaps the cleanest evidence came from Manchester City themselves.

A few months after De Bruyne signed his new contract, City paid around €118M for Jack Grealish — one of the players my model identifies as relatively close to De Bruyne’s creative profile. Grealish was still a step down in output. City themselves had therefore provided a useful market price for acquiring a high-level De Bruyne-type player.

There was no cheap version of Kevin De Bruyne. And once transfer fees enter the calculation, paying the existing player another few million a year starts looking very different.

What this gave De Bruyne: City could save on salary only by exposing themselves to a much larger replacement cost.

Tool #3: Map his outside options

Who could actually afford him — and who would need him?

Replacement cost tells City what happens if De Bruyne leaves. But De Bruyne also needs a credible answer to another question:

Where could he go?

The case describes Analytics FC approaching this in layers.

• First: affordability.

• Then: tactical fit.

• Then: squad composition.

A rich club with a 24-year-old star already occupying your role is not necessarily an attractive destination. A rich club whose chief creator is 32 may be.

So instead of asking which clubs liked De Bruyne, the analysis asks which clubs could both pay him and plausibly need him.

We can reconstruct part of that.

Assume a target salary of roughly €20M annually and use a simple wage-structure rule: one player should not consume much more than 10% of the club’s total wage bill. Only six clubs in Europe’s top five leagues at the time cleared that threshold: Barcelona, Real Madrid, Paris Saint-Germain, Manchester United, Bayern Munich; and Juventus.

Dozens of famous clubs could not pay him without materially distorting their wage structure. Then look at the chief creators at those six clubs in 2020/21.

Manchester United had a 25-year-old Bruno Fernandes. Bayern Munich had a 24-year-old Leroy Sané. Neither had an obvious succession problem.

The remaining four did.

Barcelona, Real Madrid, PSG and Juventus were relying heavily on creators aged roughly 30 to 33. All four had the finances to pay De Bruyne. All four were approaching a point where succession planning in creative positions would become increasingly important.

De Bruyne did not need to walk into negotiations claiming:

Real Madrid are going to sign me.

He only needed to demonstrate that the outside market was real.

What this gave De Bruyne: evidence that the small group of clubs capable of matching his salary included several with a plausible need for exactly his type of player.

Tool #4: Turn De Bruyne into club revenue

What was he worth specifically to Manchester City?

Now the negotiation changes. So far, we have established that:

• De Bruyne was producing at an extreme level;

• replacing him would be expensive;

• other elite clubs could plausibly afford him.

But Manchester City already knew De Bruyne was brilliant. Walking into a boardroom with another chart proving it does not necessarily change the offer. The more interesting method described in the HBS case converts football performance into something the board budgets directly: money.

The basic idea is simple.

Take Manchester City. Remove Kevin De Bruyne. Replace his minutes with a more ordinary player. Then simulate the season again and again.

How often do City still qualify for the Champions League? How often do they win the league? How much money changes hands when those probabilities move?

That is no longer an abstract player valuation. It is an estimate of what De Bruyne is worth to this particular club.

Price the player by deleting him

I rebuilt a simplified version of the chain.

Link 1: Player output → goals

For this link I use De Bruyne's 2020/21 rate — 0.67 expected goal contributions per 90, rather than his sustained 0.75 — because it is that specific season we are about to re-run. The median attacking player produces around 0.25. Over the equivalent of roughly 22 full matches, swapping De Bruyne for the median attacker removes around nine expected goals from City's season.

Link 2: Goals → points

Across eight completed seasons in Europe's top five leagues — 760 team-seasons — league points and goal difference have an extremely strong relationship. The relationship explains around 94% of the variation. On average, one additional goal of goal difference is associated with roughly 0.63 league points. So nine lost goals translate into approximately six lost points.

Link 3: Points → Champions League probability

Six points still sounds abstract. What matters is whether those points change the outcome of the season. So I simulated it.

Start with City’s actual 86 points in 2020/21. In one scenario, leave them untouched. In the other, remove the six points associated with replacing De Bruyne.

Then add random variation of roughly ±4 points — the typical gap, measured across those same 760 team-seasons, between a team’s actual points and what its goal difference predicts. That gap is the part of real football results the model can’t explain: close-game luck, red cards, deflections and everything else that moves seasons around.

Each simulation also draws a different Premier League fourth-place threshold from historical seasons. Sometimes Champions League qualification requires 66 points. Sometimes 75. Run 10,000 simulated seasons in each scenario and count how often City finish above the line.

The result:

• With De Bruyne: 99.8% Champions League qualification probability.

• With a median attacker replacing him: 96.9%.

Roughly a 3 percentage-point difference.

The HBS case reports Analytics FC obtaining a figure of approximately five percentage points from this kind of analysis. Given the differences between their internal model and my public-data reconstruction, getting into roughly the same range is encouraging. Their model was considerably richer. Mine uses season-level totals and a simplified goals-to-points bridge. They also ran their analysis while the season was still happening. I have the unfair advantage of knowing City eventually finished on 86 points.

Three percentage points of what?

This is where the exercise becomes more interesting. According to the case, Champions League qualification was worth at least €76M. Multiply that by three percentage points and you get only around: €2.3M per year.

That does not justify a €20M salary. So did the signature argument fail?

Not really.

The number is small because Manchester City were too good in 2020/21 for De Bruyne’s qualification value to be large.

They won the league by twelve points. Remove six points and they are still comfortably inside the Champions League places.

Now run exactly the same exercise for a 70-point team sitting near the qualification boundary. The result changes dramatically. Deleting those nine expected goals moves Champions League qualification probability from roughly 53% to 15%. That is a 38-point swing.

Multiply that by €76M and the same player’s contribution is suddenly worth around €29M a year from Champions League qualification alone.

Same player. Same performance. Completely different financial value. That gives us perhaps the most important idea in the entire case:

Player value is not just a property of the player. It is a property of the player-club combination.

And Manchester City’s own history shows the same thing.

Two seasons earlier, City won the Premier League with 98 points. Liverpool finished on 97. In a race that tight, a six-point player is not merely useful. Run the same simulation around that margin and City’s title probability falls from around 56% to 22% when those points disappear.

That is the difference between something close to a coin flip and something closer to a long shot.

And Champions League qualification is only one revenue line. In 2020/21, reaching the Champions League last 16 added around €9.5M. The quarter-final added roughly €10.5M. The semi-final another €12M. The final another €15M. That is approximately €47M of progression money beyond qualification, before coefficient and broadcast-pool payments.

City reached their first Champions League final that season, with De Bruyne captaining the team during the run. Then add Premier League merit payments, domestic competitions and the commercial value associated with a franchise player.

No single number needs to justify the whole contract. The point is to change the unit of the negotiation. The conversation is no longer:

What does a comparable midfielder earn?

It becomes:

What does Manchester City risk losing if Kevin De Bruyne is not here?

That is a much more favourable question for Kevin De Bruyne.

The negotiation pack, assembled

Put the four tools together and the case becomes much clearer.

1. Market benchmark : De Bruyne’s performance sat around the 98th percentile of elite European attackers. His wage sat around the 98.5th. He was paid like an exceptional player because he produced like one.

2. Replacement cost : The true substitutes were either superstars themselves or carried enormous transfer costs. A few months later, Manchester City paid €118M for Jack Grealish. Keeping De Bruyne was expensive. Replacing him could be far more expensive.

3. Outside options : Only a handful of European clubs could realistically afford his target salary. Several of them were approaching obvious creative succession problems. His alternatives were limited — but credible.

4. Club-specific value : Removing De Bruyne from City reduced the probability of financially important outcomes. More importantly, the exercise changed the frame of the negotiation from salary benchmarking to financial value and replacement risk.

The data did not need to prove that Kevin De Bruyne was good. Everyone in the room already knew that. It needed to give De Bruyne a way to negotiate against a club that possessed more information, more experience and more institutional power than he did.

Instead of saying “I deserve more because I am one of the best players in the world”, the analysis allowed him to say something closer to:

• Here is what replacing me costs.

• Here is who else could afford me.

• Here is what my presence does to your probability of winning things.

• And here is what those outcomes are worth.

In April 2021, Manchester City agreed a new deal running to 2025, reportedly worth around €80M.

According to the case study, De Bruyne had entered negotiations facing an offer that would reduce his pay. And yet, he left with a substantial raise.

Boom — that was my take on Kevin De Bruyne’s use of data scientists.

Kevin De Bruyne’s use of data is another perfect example of High-Skilled Work in practice: combining performance metrics with financial data to move beyond describing how good someone is and towards quantifying what that performance is actually worth.

In this case, that meant linking football output to wages, replacement costs, outside options and ultimately the financial value De Bruyne created for Manchester City. The interesting part is not any single model, but how different analytical tools can be combined to answer a real commercial question.

The original Analytics FC report clearly contained more analysis than I could reproduce here from public data. It goes without saying that this piece is a reconstruction of the main ideas described in the case study, not a replica of the full report.

If you want to go deeper into the project itself, I’d highly recommend Chris Gill’s interview with Analytics FC’s Jeremy Steele, as well as the original Harvard Business School case study. Both are brilliant.

I hope you enjoyed reading this deep dive as much as I enjoyed putting it together.

Subscribe now

Thanks for reading all the way to the end.

Talk soon,

Martin

P.S. This post was AI-assisted for editing and code debugging. Kevin De Bruyne’s contract was not. But maybe the next one will be.

The clip is an excerpt from A Beautiful Obsession, Amazon Prime’s documentary on the club. In it, City’s process for replacing Kevin De Bruyne is described as analysts taking a large set of metrics and melting 43 of them down into 7.

A lot of the replies were asking roughly the same thing: why would City show this publicly?

The answer is that they haven’t really given away the secret sauce. We don’t know their exact variables, their data or how the final ranking is constructed. But the statistical idea behind what they’re describing is actually pretty standard.

Which got me thinking: with the data I have, how close could I get?

So that’s what I did. I took eight seasons of data on every player in Europe’s top five leagues, built a statistical fingerprint of peak De Bruyne, compressed 43 metrics into seven dimensions, and asked the data a simple question:

Which player in 2024/25 looked most like peak Kevin De Bruyne?

The model was allowed to answer whatever it wanted.

Out of 673 players, it answered: Rayan Cherki.

The player Manchester City actually signed.

Here’s what I’m going to do: explain what those “43 metrics into 7” actually means, rebuild the process in R, see why Cherki comes out on top, and then look at where the model stops and human scouting takes over.

1. The problem: how do you replace Kevin De Bruyne?

First, some context on the professional problem, because “replace Kevin De Bruyne” is a genuinely hard one.

De Bruyne left City in the summer of 2025 after ten years. For most of that decade he was the best creative midfielder in the world. You cannot walk into a shop and buy another one.

What a modern recruitment department can do is search systematically. Clubs like City sit on enormous databases: every pass, carry, shot and duel, for tens of thousands of players, going back years.

That creates the opposite problem. Too much information.

A single season of one player’s data contains dozens of plausible metrics. Key passes. Expected assists. Progressive carries. Through balls. Which ones matter? How do you compare a player who is elite on twelve metrics with one who is elite on fifteen different ones?

The clip’s answer: you don’t compare 43 numbers. You compress them into seven, and compare those.

That compression step has a name.

2. How to turn 43 metrics into 7

The technique the clip describes is, in all likelihood, some form of dimensionality reduction. The classic version, and the one I will use, is called Principal Component Analysis, or PCA.

Think of each metric as a question you ask about a player. How often does he pass into the box? How often does he carry the ball forward? How many chances does he create?

Ask 43 questions and you will notice something: many answers move together. A player with lots of key passes usually has high expected assists too, and plenty of shot-creating actions. Those are three questions, but they are mostly measuring one underlying trait: this player creates chances.

In other words, 43 metrics do not contain 43 independent pieces of information. The real information lives in a handful of underlying traits, and each metric is a noisy reading of one or more of them.

PCA is the algorithm that finds those traits automatically. It looks at which metrics move together across hundreds of players and builds new, compressed variables called components. Each component is a weighted blend of the original metrics, and the first few components capture most of the information in the whole table.

Or in plain English: PCA turns 43 (possibly) correlated columns into a few uncorrelated ones while losing as little information as possible.

The clip says City’s analysts ended up with seven. So seven is what I will use.

3. Rebuilding City’s search

Before doing that, one obvious limitation. I do not have City’s data, their 43 metrics, their preprocessing, or their similarity function. Nobody outside the club does. What I can rebuild is the type of workflow the clip describes, using data available outside club walls. The purpose is to understand the method, not to claim access to City’s model.

So I fired up R.

The data. My dataset is Opta event data, aggregated to season-level statistics. It covers the top five European leagues across eight seasons, 2017/18 to 2024/25. Roughly 2,700 player-seasons per year.

The target. I defined “peak De Bruyne” as his five seasons with at least 20 full matches played: 2017/18, 2019/20, 2020/21, 2021/22 and 2022/23. Pooled together, that is about 139 matches of football, converted into per-90 rates. This deliberately excludes his injury-wrecked 2018/19 and his final two declining seasons.

The 43 metrics. I picked 43 variables covering everything the data can see about an attacking player. All are per-90 rates or percentages. The full list, grouped:

• Passing (12): key passes, expected assists, expected assisted goals, passes into the penalty area, crosses into the penalty area, passes into the final third, progressive passes, progressive pass distance, pass attempts, pass completion %, long-ball attempts, long-ball completion %.

• Pass types (3): through balls, switches of play, crosses.

• Shooting (5): non-penalty xG, shots, shots on target %, xG per shot, average shot distance.

• Chance creation (5): shot-creating actions, goal-creating actions, and shot-creating actions arising from live passes, from take-ons and from shots.

• On the ball (13): touches, attacking-third touches, penalty-box touches, take-ons attempted, take-on success %, carries, progressive carry distance, progressive carries, carries into the final third, carries into the box, ball losses, passes received, progressive passes received.

• Defending (2): tackles plus interceptions, attacking-third tackles.

• Duels and pressure (3): fouls drawn, ball recoveries, aerial duels won %.

Why 43? Honestly: because the clip said 43. It is an editorial choice matching the narrative, not a claim that these are City’s variables.

The comparison pool. Every midfielder and forward in the top five leagues in 2024/25 with at least 15 full matches. That is 673 players.

The compression. Standardize each metric so no variable dominates just because of its units, then run PCA. The first seven components retain about 80% of the variance in the original 43 columns. In other words, seven numbers per player preserve roughly four-fifths of the information contained in the original 43. That is the “melting.”

What the seven dimensions are. Each component groups metrics that tend to move together across the 673 players. Looking at which metrics load most heavily on each one, here is my reading of what the algorithm found:

The labels in the last column are my interpretation of the loadings, and a different analyst might word them differently.

The search. Project peak De Bruyne into this seven-dimensional space, then measure the straight-line distance between him and each of the 673 players. Smallest distance = most similar statistical profile.

And that's basically the process described in that one line of the clip.

4. The closest match: Rayan Cherki

Here is the top of the leaderboard.

Out of 673 players, a 20-year-old at Lyon had the closest statistical profile to peak De Bruyne in Europe, in the exact season before City signed him.

What does a distance of 4.2 actually mean? On its own, nothing: the units are abstract. It becomes meaningful with reference points. The median player in this pool sits at a distance of about 15 from peak De Bruyne. Pick two attacking players at random and they are typically about 7.5 apart from each other. So Cherki’s 4.2 means he is closer to peak De Bruyne than two ordinary attacking players are to one another. Only five players in Europe came within a distance of 6.

And the names immediately behind him — Olise, Wirtz, Ødegaard and Bruno Fernandes — are reassuring. The model is clearly picking up the type of creative profile we would expect.

Where does Cherki actually match De Bruyne, metric by metric? The chart below compares them in z-scores: how many standard deviations a player sits above or below the average attacking player, where zero is exactly average, +1 is better than roughly 84% of the pool, and anything beyond +2 is elite territory.

On chance creation, they match almost perfectly: key passes, expected assists, passes into the box and through balls are all three-to-five standard deviations above the average attacking player, for both of them. Where they differ is the delivery mechanism.

Peak De Bruyne got the ball into the box with crosses and raking switches of play. Cherki does it off the dribble, with more take-ons and more touches inside the box, and does less defensive work than even De Bruyne’s famously modest amount.

Same creative output. Different route to it. That’s an important nuance and we will come back to it.

5. Did I just get lucky?

Cherki coming first this cleanly made me suspicious. There are plenty of choices in this analysis that could affect the ranking, so I reran it under several different specifications.

Number of components: with 3, 5, 7, 10 or 15 components, Cherki is first every time.

Minutes threshold: at 10, 15 or 20 full matches minimum, Cherki is first.

Definition of peak: if I scrap the “peak seasons” choice entirely and pool all eight De Bruyne seasons, injury year and decline included, Cherki is still first, and the top ten barely reshuffles. The two versions of the reference profile turn out to sit almost on top of each other, mostly because the peak seasons already contain three-quarters of his career minutes.

Overall, in this data, under this family of specifications, Cherki’s 2024/25 season simply is the closest thing to peak De Bruyne that existed in Europe’s top five leagues.

6. What the model cannot see

This obviously doesn't mean City's decision was made by an algorithm. If anything, the clip shows the opposite: the model appears to have been a way of narrowing the search before the much harder human work began.

A similarity score is where recruitment starts, not where it ends. My model knows nothing about:

• Physical and tracking data. Distances, sprints, off-ball movement. City has it; season-level event data does not.

• League context. Cherki’s numbers were produced in Ligue 1, De Bruyne’s in the Premier League. A per-90 profile does not automatically translate across leagues: opposition strength, tempo and tactical norms differ, and a club would adjust for that.

• Tactical fit. Cherki played as a free-roaming creator at Lyon. Can he play in City’s positional structure? A statistical profile cannot answer that. Scouts watching hundreds of hours of footage likely can.

• Mentality and references. Cherki had a reputation as a brilliant but inconsistent talent. Clubs interview coaches, teammates, staff (as the clip shows). None of that is in any dataset.

Overall, the metrics say Cherki creates like peak De Bruyne, but through take-ons rather than crosses and switches. A naive reading says “perfect replacement.” A professional reading says “elite creator with a different delivery mechanism, who will need a different tactical environment.” Maybe that’s why the analysts expected Cherki to reach peak De Bruyne level (and style?) in two years time.

7. The model finds the shortlist. Humans make the decision.

City aren’t using the model to make the decision. They’re using it to make the search manageable. Start with thousands of players, reduce the information into something you can compare, produce a shortlist, and then spend human time investigating the players who survive that process.

That structure is not specific to football. It is just the example here. A huge part of high-skilled analytical work is exactly this: use data to shrink the problem, then spend expensive human judgment where it matters.

The viral clip made 7.5 million people feel they had glimpsed something super confidential inside Manchester City.

And yet, there isn't really a magic model. There is a fairly standard statistical tool sitting inside a much larger recruitment process.

So, did we recreate Manchester City’s search?

Obviously, this isn’t City’s model. But with previously (😅) public data, PCA and about five hundred lines of R code, one could reproduce the idea described in the clip — and the model landed on the same player (😮‍💨).

What I couldn’t reproduce is probably the more valuable part: everything City knew about Cherki that isn’t sitting in my dataset.

Boom — that was the behind-the-scenes look at the tool elite clubs use to shortlist players.

Thank you for reading all the way to the end. I hope you enjoyed reading it as much as I enjoyed creating it.

What better way to spend a summer vacation, right? And the Premier League (❤️) starts in a couple of hours. What a great day.

Talk soon,

Martin

P.S. This post was AI-assisted (editing and code debugging). But not all em dashes were, I promise.