Welcome to the first Moneyball Squad Audit, where I’ll ask one question about Man United: is the money in the squad doing its job?

I’ll look at the team in three ways:

• what the players are worth on the market,

• what they are paid, and

• how the team performs.

I’ll then put the three next to each other, against the rest of the league, and against the club’s own past.

For United that means nine completed seasons, from 2017/18 to 2025/26, plus the first five rounds of the current 2026/2027. Over that period the club spent about €3.9 billion on wages and €2.2 billion on transfer fees.

The full report runs to 86 pages and more than 21,000 words, with 59 figures. You can download it as a PDF at the end of this post. If you would rather not read all of it, here is an extended summary of what I found.

Three rules throughout. Every amount is in euros, and past transfer fees are restated in 2026 euros, so a fee paid in 2017 counts for what that money would buy today. Whenever seasons are compared, each club is measured against the league average of the same season, which removes inflation and the changes in data sources at once.

Let’s get into it.

1. Did United’s money buy results?

Start with the plain benchmark. Over the eight seasons with published accounts, 2017/18 to 2024/25, the clubs that won the league paid on average 2 times the league’s average wage bill and took 2.43 points per game. Clubs finishing 2nd to 4th paid 1.7 times and took 1.99.

United paid 1.86 times the league average. Almost the bill of a champion. But it took 1.72 points per game, or about 65 points a season: the return of a club finishing 5th-6th.

That gap is not the product of one bad season. Drawing the league’s usual return as a line, United sat below it in six of nine seasons. The worst was 2024/25, 0.66 points per game below the line, when the league’s 5th-largest wage bill finished 15th. In 2021/22 the largest bill in the league finished 6th.

Squad value tells the same story. United fielded one of the six most valuable squads in every season, yet fell below the value line in five of nine. Relative to the league, its squad slid from 2.1 times the average value in 2017/18 to 1.4 times today, because other clubs grew faster.

Because the two measures agree, the explanation is unlikely to be that United’s players were overpaid for their value, or overvalued for their wages. However the squad is priced, it delivered less than squads of that price usually do. The shortfall recurs under different coaches and rebuilt squads, which makes bad luck an unlikely explanation and points instead to how the money was used.

Subscribe now

2. Where did the performance fall short?

The obvious suspect is the attack.

Setting each club’s goals against the market value of the players who produced them, United’s front line scored about 22 goals fewer than attacks of its value usually score in 2023/24, and 23 fewer in 2024/25. The defence, by contrast, has conceded about what a defence of its value suggests since 2022/23.

It looks like a finishing problem. It was not.

Was the shortfall in the chances or in the finishing?

In the chances. United made fewer chances, created minus conceded, than its wage bill usually buys in all eight seasons with published accounts. The worst was 2023/24: 1.09 expected goals per game short of the line, about 41 over a season. 2025/26 was the first season of the period on the line.

The attackers finished what they did make at about par: 25 non-penalty goals from 28 expected in 2023/24, and 25 from 25 in 2024/25. They were not wasting chances. They were not making enough of them.

If anything, United usually got more than its chances deserved. In seven of nine seasons it took more points than those chances usually bring.

The finishing, the goalkeeping and the margins of close games did not cause the shortfall; they covered for part of it.

How much did luck flatter the table?

To find out, I replayed every match of every season 10,000 times from the shots both teams took, each shot going in with the probability its expected goals give. The idea is simple: keep the chances, re-roll the finishing, and see where the team usually ends up.

In 2017/18 United finished 2nd. Its median replay was 6th, and only 1% of replays reached 2nd or higher (I guess Mourinho was right to claim that finishing second with United that season was “one of the best achievements in his career” after all). The side that finished 8th in 2023/24 had, on its chances, the profile of a 15th-placed team. Over the nine seasons United took about 47 points more than its replays average.

Much of that cushion had a name. In 2017/18 David de Gea prevented 12.5 goals more than an average keeper would have conceded from the shots he faced, the best in the league and worth about eight points. No United keeper since has stood out from what luck alone would produce.

The lesson is uncomfortable for anyone who reads the final rankings table first. In its best seasons the table showed a better team than the chances did. A club that reads its points rather than its chances is likely to have drawn the wrong lessons from those seasons.

Who created the chances that did exist?

Mostly one player. Since 2020/21 Bruno Fernandes has made between a fifth and a third of United’s expected assisted goals in every season, 35% last season. At the median club, the top creator makes 16% to 20%.

A team that creates through one player has little to fall back on. That dependence is also the other side of the captain’s value, and one reason his contract, which ends in June 2027, carries so much weight.

How much did injuries take?

More than at most clubs, but mostly through numbers rather than price. In every one of the nine seasons a larger share of United’s squad was out injured on an average day than at the median club. Weighted by market value, 28% of the squad’s value was missing on an average day of 2019/20, against a league median of 17%, and about a fifth in 2023/24 and 2024/25.

Counting each injured player’s contractual wage for the days he was out, United paid about €345M to players while they were injured over the nine seasons, more than any other club. Those are wages the club owed in any case, not extra spending caused by injuries, and United’s wage bill is large. As a share of its own bill the figure is 16%, 4th of the 19 clubs with five or more seasons in the league, against 14% at the median club.

Whether the injuries caused the bad results or the same squad problems caused both, the data cannot say. Three of the four heaviest injury seasons were among United’s worst for money; 2019/20, the heaviest of all, was not. That suggests the injuries deepened bad seasons rather than causing the long-run shortfall.

3. Where did the money go?

The second half of the explanation is what happened to the money. Since 2017/18 United has spent €2.2 billion on transfer fees, in 2026 euros, and received €610M back: the largest net spend in the league.

Did United overpay to buy?

No. When buying, United paid 1.20 times the players’ market value at the time, against 1.28 across the league: the premium a buyer pays for the players it wants was normal. The difference is at the exit. When selling, United received 0.94 times market value, where the league as a whole received 1.13, and it sold below value in six of ten seasons.

What came back from the signings?

To answer that, I followed every signing to its outcome: the fee received if the player was sold, nothing if he left for nothing, and his market value today if he is still at the club. Run for every club with ten or more signings of €5M or more, it gives a simple ratio of what came back for each euro of fees.

United got back €0.48 for every euro. The median club got back €0.85. United ranks 26th of 27; only Sheffield United did worse.

A raw ledger is not a verdict on its own. Players age, and a fee is rarely recovered five years later, so I set each signing against what signings of the same age, held for as long, usually bring back across the league. United still gets 69% of the usual, still 26th of 27.

The detail is the interesting part. The players still at the club are close to what age and time alone would leave (89% of the usual). The players United sold fetched about half of what similar players usually fetch (54%), and several left for nothing. That points to selling late and from weakness, players the club no longer wanted, more than to paying too much in the first place.

Did the coaching changes play a part?

United has had six coaching regimes since August 2017. That is the league median, but the clubs that win almost never change mid-season: Manchester City had two regimes in the period and Arsenal three.

Across the league, a signing made under a coach who left within a year brings back 42% of his fee at the median, against 63% when the coach stays. 51 of United’s 55 signings since 2017 were made by coaches who have since gone.

Clubs sack coaches because seasons go badly, so this is an association, not a measured effect. The mechanism is plausible, though: each new coach buys for his plan, and each departure leaves players recruited for a plan that no longer exists. The result is a squad whose value fell over the year in seven of nine seasons, by 5% a year on average, while the median club’s squad gained 4%. Much of the transfer budget went on replacing value the squad had lost.

4. What does the squad look like today?

Young value, old wages. United’s squad is the 5th most valuable in the league, at €929M, and its wage bill the 3rd largest, at about €274M a year. The value is young and sits in the attack, much of it bought in the last two windows.

The wages are older and sit in the midfield, the defence and a senior core whose contracts were set when their value was much higher.

The single number that sums it up: United holds €3.40 of squad value for every euro of annual wages. The typical club holds €4.80, and Arsenal, with a similar bill, €5.60. United ranks 18th of 20. Its four best-paid players take 28% of the wage bill and hold 13% of the squad’s value.

Figure 11. Squad market value against the annual wage bill, 20 clubs of 2026/27.

The next big decision is close. The contracts ending in June 2027, Fernandes, Martínez, Maguire and Shaw, carry 23% of the wage bill, about €62M a year, and 11% of the squad’s value. Three of the four are defenders, in the line the club pays furthest above its value.

Not every expensive contract is poor value. Measured by the chances he made, Fernandes added 8.5 expected points in 2025/26 above a replacement-level player, the most in the Premier League, at €2.5M of wages per point against a league median of €5.1M. With him on the pitch, United’s expected-goal difference was almost a goal per game better than without him over the last three seasons.

One caution on individual players. The performance index in the report measures what a player did in his team’s system that season. For a young player in a new side it is the least reliable number in the report, and a low score is not evidence that a transfer was a mistake or that a player lacks potential.

5. What would show that United has turned?

The audit cannot say whether the improvement of 2025/26 will last, and five rounds into this season United sits 12th with 5 points. What it can say is what to watch.

• The first sign would be chances: a season in which United again creates what its wage bill usually buys, as it did last season for the first time in the period, without relying on points above its chances to get there.

• The second would be the ledger: signings that hold their value, and players sold at or above their market value rather than below it.

• The third would be stability: a coach who stays long enough for the players he signs to be worth what they cost when he goes.

None of these is a statistical question. The data shows where the money went and what it bought; what the club does with that is a judgement for the people who run it.

Boom — that was the Manchester United Moneyball Squad Audit, 2017/18 to 2026/27.

Hopefully you now have a clearer idea of why United underperformed.

If you only want to remember a few things, I’d pick these:

• For years, United spent like a champion on wages but got the points of a club fighting for 5th place. United finished about 10 points a season short of what its wage bill usually buys (which is around 78 points over eight seasons).

• United’s transfer problem is mostly at the exit. When it sells, it gets 0.94 times a player’s market value, against 1.13 for the league as a whole.

• The players United bought for today’s squad are worth 70% of what they cost, the lowest in the league. The median club’s are worth 111%.

• Over nine seasons, United’s squad lost 5% of its value a year on average, while the median club’s gained 4%.

• United’s attack wasn’t wasteful: it finished its chances at about the rate they deserved. It just didn’t create enough of them for what it costs.

• The defence concedes like a mid-table defence but is paid like a top-four one.

If you’re interested in the full report, you can download it below.

Download the full Moneyball Squad Audit

I hope you enjoyed this one as much as I enjoyed putting it together.

What’s the next team you’d like me to moneyball audit next? Drop your suggestions below.

Thank you for reading until the end ❤️.

See you next week,

Martin

PS. If you liked this piece chances are you’ll also enjoy how Kevin De Bruyne himself used data scientists to secure a 30% increase at 29, or how Man City used a magical data science technique to find his replacement or what Bruno Fernandes’ data science team likely argued in his ongoing negotiations with United.

P.P.S. If a tailor-made report (such as the one I did on Kai Havertz’s contract negotiation) is what interests you (including on women’s football), drop me a line.

He had broken into the first team, impressed against Real Madrid in a Champions League quarter-final and scored on his England debut, becoming the youngest player to score on his senior England debut.

The season before, his salary was estimated at £4,000 a week. After the deal? Capology puts him on about £45,000. Valuball estimates £80,000. Neither is a confirmed contract figure, but even the lower estimate implies an enormous 10x raise.

To secure his deal, he brought in data scientists — the data consultancy that had also advised Kevin De Bruyne on his €80M contract. But unlike the greatest modern playmaker, he faced a very different objection from the club: you’re 18, and we’ve only seen one season.

ESPN has described two analyses used in his negotiations:

• a comparison with other teenage breakouts

• and research into the wage gap faced by academy graduates.

The full private report is of course not public.

But today I’ll reconstruct those two arguments in an attempt to uncover what (likely) happened on the negotiation table.

Here is what I found.

1. How rare is a teenager who actually plays?

Arsenal could reasonably say Lewis-Skelly had only one good season. His advisers needed to show just how unusual that season was.

ESPN reported that Analytics FC identified 69 players who made their top-five-league debut from 2017/18 and played at least 600 minutes that season. I rebuilt a comparable group: first season in one of Europe’s five biggest leagues, 18 or younger, at least 600 league minutes, and no earlier top-five appearance going back to 2014/15.

My count is 106 players across eight seasons, not 69. The exact filtering behind the private exercise is not available, so I won’t pretend these are identical cohorts. But they tell the same story:

Teenagers who enter an elite league and immediately hold a place are not common.

The list includes Vinícius Júnior, Bellingham, Pedri and Gavi. It also includes plenty of players whose careers went nowhere near those heights. In other words, rarity is not the same as future stardom.

Lewis-Skelly played 1,369 league minutes as a 17-year-old. On my full-back performance index, he was at the 75th percentile among top-five-league full-backs and ranked 34th of the 104 players in the cohort I could assess.

Not Pedri. Not some once-in-a-generation outlier. But an academy player who had already done something most teenagers never manage.

2. The pay rise looks different when you compare peers

Now the money. Before his renewal, Lewis-Skelly’s estimated wage sat at the 9th percentile of the Premier League. A player starting in the Champions League against Real Madrid was being paid like the bottom tenth of the division.

After renewal, the estimate moves him to the 58th percentile. That is a jump of 49 percentage points. The typical player in my teenage-breakout group moved from the 22nd to the 39th: 17 points.

By that comparison, Lewis-Skelly got an unusually large correction. But he did not suddenly become one of the league’s highest earners. He moved from the bottom to a little above the middle.

And there is a second, more uncomfortable comparison for Arsenal:

What were other Premier League full-backs getting paid to do his job?

Among Premier League full-backs with at least 900 minutes in 2024/25, Lewis-Skelly sat at the 75th percentile on my performance index. The median player sat at the 71st. Yet Lewis-Skelly’s estimated new wage was £80,000, against a median of £90,000.

A fitted market curve puts a full-back at his measured level at roughly £91,000 a week. On Valuball’s estimate, he is on about 88% of that benchmark. Of the 33 full-backs paid at least as much, 17 scored lower on my index.

So he was above the middle of the league on measured performance and below it on pay. Was Arsenal getting a discount? Possibly. But before we call it underpayment, there are two obvious complications: his age and his academy contract.

3. Being 18 has a price

How much do clubs pay for performance when the player is barely old enough to sign the contract?

I modelled wages across 8,090 player-seasons in Europe’s top five leagues, allowing for prior performance, club wage bill, position, league and season—but deliberately leaving age out. I then looked at how the gap between actual pay and predicted pay varied with age.

The pattern is stark.

Players aged 17–19 earn about 62% below the quality-based benchmark on average.

At 20–21, the shortfall is 47%; at 22–23, 25%. By 24–26, it largely disappears. Older players tend to earn a premium.

In other words, clubs do not pay purely for what the current season says. They also pay for evidence that you can keep doing it.

For Lewis-Skelly, the no-age benchmark at Arsenal’s wage level was about £111,000 a week before renewal. He was estimated to earn £4,000: 96% below. Using the following season’s wage data and provider, the equivalent benchmark is about £159,000; £80,000 leaves him 50% below it. These are separately estimated benchmarks, not an apples-to-apples jump from £111,000 to £159,000.

That sounds brutal—until I put age back into the model. His new wage then comes out only about 7% below the expected amount for a player of his age, measured quality and club.

That is the surprise. On the published £80,000 estimate, Arsenal did not necessarily give him a superstar premium. It appears to have brought him roughly up to the market rate for an 18-year-old like him.

A big raise and a generous contract are not automatically the same thing.

Subscribe now

4. Does coming through the academy cost you even more?

This is the piece of the story I find most interesting. The age discount is widespread; the academy question is about the club that already owns your registration and helped develop you.

ESPN reported that Analytics FC revisited research first prepared for Mason Mount’s Chelsea negotiations. A Harvard Business School case on the consultancy describes an even bigger claim in an academy-player case:

Players who stay may earn roughly a third of what they could earn after moving.

That implies a 67% gap.

I wanted to see whether anything like that survives a broader test.

I compared players within the same club and season, controlling for age, position and measured quality. Players the club paid a transfer fee for are classified as signings; players who broke through at the same club as teenagers and were never bought are classified as academy products. Unclear cases are left out.

Across 7,812 player-seasons and 675 club-seasons, academy graduates earned about 45% less than comparable team-mates signed from elsewhere.

Not the 67% described in the case study. Still a very large gap.

And it is concentrated in the years that mattered to Lewis-Skelly. The estimated gap is 48% at ages 17–19, 49% at 20–21, and 37% at 22–23. By 24–26, it falls to 8%, with too much uncertainty to distinguish it reliably from zero.

I tested nine versions of the comparison—different controls, a Premier League-only sample, different academy definitions and shorter time windows. The full-sample result stayed between 37% and 48%. The separate estimate for players aged 24 and over was about 3%, again too uncertain to call a gap.

Now the caveat: 45% is an association, not proof that loyalty causes a 45% pay cut. Clubs may pay transfer fees for qualities my index misses, then pay those recruits more. Nor can you simply add this 45% to the age discount; the two analyses answer different questions.

What would I take to the negotiating table, then? Not a claim that Arsenal owed him exactly 45% more. I would take the age profile: why should the largest gap between academy players and recruits appear precisely when young graduates have the least negotiating leverage?

That makes the argument much harder to dismiss as an 18-year-old asking to be paid like a senior star.

5. Arsenal had alternatives. Arteta kept picking him.

For a club, the simplest reply to a wage demand is “we have other players.” Arsenal certainly did. At left-back they could turn to Riccardo Calafiori, Jurriën Timber, Oleksandr Zinchenko or Kieran Tierney.

So I looked at the minutes rather than the names on the squad list.

Lewis-Skelly had played just 22 Premier League minutes at left-back before December 2024. From December onwards, he took 54% of Arsenal’s minutes in that position. By May, his monthly share was close to four in five.

Arsenal could say: we have four other options. His representatives could answer: yes—and the manager kept choosing the teenager.

I know which of those two lines I would put on a slide.

But being important to Arsenal is only half a negotiation. What happens if another club comes for him?

ESPN reported interest from Real Madrid. That does not establish a formal offer or tell us how serious the interest was. But it does give us a testable question: would his game actually fit Madrid’s left-back role?

I compared 15 aspects of left-back play across 90 top-five-league clubs in 2024/25, from passing and progression to defending and crossing. Arsenal was the closest match—unsurprisingly, since Lewis-Skelly contributed to its profile. Real Madrid ranked second.

Madrid also had the largest wage bill in the sample. Fran García, 26, played 68% of their left-back minutes that season; Ferland Mendy, 30, took most of the remainder.

So the reported interest was not obviously a tactical mismatch. That is evidence of a plausible destination, not evidence of an actual bid. The latter would matter far more at a negotiating table.

6. The chart Arsenal should have brought

So far, the case for a substantial raise looks strong: rare teenage minutes, good measured performances, pay near the bottom of the league, and a manager who trusted him.

But the best argument against him comes from the same group of teenage breakouts.

I followed the 63 players in that group whose careers could be observed for three seasons after their breakthrough. Here is where they ended up.

Only 35% were still regular starters or regular-minute players in a top-five league three seasons later.

Another 38% remained in a squad but played less than half the available minutes. 27% were no longer in a top-five league.

The better performers had somewhat stronger outcomes: 39% of the top half became regulars, versus 31% of the bottom half. But a breakout season was nowhere near a guarantee.

If I were on Arsenal’s side, I would point at that chart and ask: why should we pay today for a ten-year career when two-thirds of the comparison group were not regulars three years later?

That is a fair counter-argument. Lewis-Skelly had shown that he could play at the level. Arsenal still had to price the risk that he would not maintain it.

7. What happened after he signed?

Any fair assessment also needs the benefit—and the danger—of hindsight. These numbers were not available when he signed in July 2025.

His share of Arsenal’s league minutes fell from 42% in 2024/25 to 24% in 2025/26. Read only that figure and you might conclude that his place disappeared.

Across all club competitions, though, he played 36 matches and 2,034 minutes in 2025/26, compared with 38 matches and 2,316 minutes the season before. Cups and Champions League appearances made up part of the league shortfall, while Calafiori took more Premier League minutes.

After a period with England’s under-21s, Lewis-Skelly returned to the senior squad in September 2026. As of 10 October 2026, he had appeared in five of Arsenal’s first six league matches, playing 64% of the available minutes. That is a return to involvement, not yet proof of a settled first-choice position.

I would call it a year of rotation, not a collapse. The longer-term bet is still being tested.

So, did the data win him the deal?

We cannot know how much of the eventual contract was caused by the data team. The discussions are private, the wages are estimates and Arsenal had reasons of its own to secure an excellent academy player until 2030.

But the reconstruction tells us something useful. At around £4,000 a week, Lewis-Skelly was paid like the bottom tenth of the Premier League. After one extraordinary breakthrough season, the reported raise moved him into the middle of the division.

Against established full-backs of similar measured quality, he still looked cheap. Against players of his own age, his new wage looked much closer to normal. And across clubs, young academy graduates tended to be paid substantially less than equally measured recruits.

The most powerful line in this negotiation was never “look how many times my salary went up.” It was “compare what I do with what players like me get paid.” And to show that having a data team by your side helps a lot.

Boom — that was how Myles Lewis-Skelly used data scientists to secure his 10x contract with Arsenal.

Thanks for reading all the way through ❤️.

I hope you enjoyed this article as much as I enjoyed putting it together.

Cheers,

Martin

PS. If you liked this piece chances are you’ll also enjoy how Kevin De Bruyne himself used data scientists to secure a 30% increase at 29, or how Man City used a magical data science technique to find his replacement or what Bruno Fernandes’ data science team likely argued in his ongoing negotiations with United.

P.P.S. If a tailor-made report (such as the one I did on Kai Havertz’s contract negotiation) is what interests you (including on women’s football), drop me a line.

This piece was AI-assisted. Claude asked for a 10x raise after editing it. I told it to be patient. It's still young.

The City Files is a series on the Manchester City verdict and the compensation claims that could follow it.

• Part 1: Manchester City’s 115 Charges, Explained in Plain English

• Part 2: The £35M Playbook: How Burnley Made Everton Pay, and How City’s Rivals Could Do the Same

• Part 3: What Arsenal Could Claim (you are here)

• Part 4: What Liverpool Could Claim

• Part 5: What Manchester United Could Claim

Hi friends,

In Part 2 of the series, we took apart the only decided case in which one Premier League club was made to pay another for breaking the financial rules: Burnley v Everton, £35M.

We ended with a six-step playbook.

1. Start from a proven breach.

2. Build the “what if” world.

3. Turn the money into points.

4. Turn the points into a league position.

5. Price what was lost.

6. Add interest.

Today I run that playbook for Arsenal in an attempt to quantify the damages amount they would be claiming.

I am going to play both sides. First what Arsenal’s economists would argue, then what City’s would say back, step by step, with the numbers.

Two warnings before we start.

The first is that this is my reconstruction. Nobody has filed anything, and I have no inside knowledge of what either club is planning.

The second matters more. The Commission found that City’s income was overstated by more than £830M over nine seasons, but the season-by-season amounts sit in appendices that have not been published. So wherever I need a number for a single season, I have had to assume one. I will say so each time, and I will show you how much the answer moves when the assumption changes.

Let’s get into it.

First question: which seasons?

A compensation claim needs a specific loss.

• “City had an unfair advantage for a decade” is a complaint.

• “City’s breach cost us this particular place, in this season, worth this much” is a claim.

So the first job is to look for the seasons where City finished directly between Arsenal and something valuable.

In the nine seasons covered by the Commission’s findings, 2009/10 to 2017/18, City finished above Arsenal seven times. But in six of those seven, it made little difference to Arsenal: they were in the top four anyway, and moving from fourth to third is worth a couple of million pounds in prize money.

One season is different.

In 2016/17 Arsenal finished fifth on 75 points. City were third on 78. Liverpool were fourth on 76. Arsenal missed the Champions League for the first time in twenty years, by one point.

Two things make this the season to build a claim on.

• It is the only one in the period where City’s position cost Arsenal a place worth tens of millions.

• And it is a season in which the Commission found City in breach of the Premier League’s spending limit, the Profitability and Sustainability Rules (PSR). That is exactly the kind of breach Everton committed in the season Burnley went down.

What about the two titles Arsenal lost to City, by five points in 2022/23 and by two in 2023/24? Those are the ones supporters care about most. But the financial findings stop at 2017/18. To claim for 2023 or 2024, Arsenal would have to argue that money disguised before 2018 was still winning matches five years later. I think that is a different (and harder) case, and I leave it aside here.

So: 2016/17, fifth place, three points behind City.

Step 1: The Proven Breach

In Burnley’s case this step was short. The league had already established that Everton overspent by £19.5M, and that number was binding.

For Arsenal, the breach is established too. What is missing is the number.

We know the total: more than £830M of income that should have been recorded as owner funding, across nine seasons. We know City breached the PSR in 2015/16, 2016/17 and 2017/18 “by a very substantial amount”. We do not know what “very substantial” means in pounds for 2016/17.

And the right number is not obvious even in principle. Remember what Everton’s £19.5M was: the amount by which its losses exceeded the limit. It was not the club’s total spending, and not its total losses. So for City, there are at least three candidates.

• The overstated income of that one season. If the £830M fell evenly across the nine seasons, that is about £92M for 2016/17. If it followed the growth of City’s reported commercial income, about £125M. (Both profiles are my assumptions.)

• The excess over the spending limit. The PSR test looks at three seasons together and allows a loss of £105M. City reported a combined profit of £30M for 2014/15 to 2016/17. Take out three seasons of overstated income and that becomes a loss of roughly £247M to £314M. The excess over £105M is then £142M to £209M. But clubs can deduct spending on the academy, the women’s team, community work and infrastructure before the test is applied. I do not know City’s deductions. If I assume £40M a year, the excess falls to somewhere between £22M and £89M.

• Everything in the three-year window. Arsenal’s most aggressive version: all the disguised funding of those three seasons built the squad that played in 2016/17. That is £277M to £344M.

So the money in play for 2016/17 could be as little as £22M or as much as £344M. That is not a satisfying answer, and it is the single biggest uncertainty in this piece. It will narrow sharply the day the appendices are published.

• What Arsenal would argue: follow Burnley. There, the whole excess over the limit, built up over four years, was counted against the final season, because “the benefit of the overspend is cumulative”.

• What City would argue: exactly what Everton’s expert argued. Spread the excess across the seasons of the assessment period, so that only a third of it lands on 2016/17. On my assumptions that would be £7M to £30M. The Commission rejected this for Everton. City would say its facts are different.

Step 2: The “What If” World

The counterfactual is the world in which City complied. What would it have done?

• What Arsenal would argue: City would have had to spend less on players. The disguised money was there to fund a squad the rules did not allow. Take it away, and the squad is weaker. This is the conclusion the Commission reached for Everton, and City’s breach is many times larger.

• What City would argue: a club with City’s commercial pull would have found the money legitimately. Real sponsors would have paid more over time; the owner could have funded infrastructure and the academy, which the rules allow; players could have been sold at a profit.

There are two problems with City’s version. The Commission found that City would have failed the PSR test in every season even if the sponsorship deals were revalued at a fair market price. And in Burnley v Everton, the test was what the club would “practically and realistically” have done, not the version most convenient for the club being sued. Everton’s owner had turned down a real offer for a player; that sank its “we would have sold someone” argument.

I think Arsenal has the better of this step. But City does not need to win it outright. Every pound it can show would have been raised legitimately is a pound that comes off the number in step 1.

Subscribe now

Step 3: Turn The Money Into Points

Now the exchange rate. How many points was each pound worth to City in 2016/17?

In Part 2 we saw that there is no observable exchange rate between pounds and points. The experts have to build one, and the two sides in Burnley’s case built it very differently. I have rebuilt both methods for City.

Arsenal’s method: City’s own rate

This is what Burnley’s experts did and the Commission accepted. Divide City’s points by its player spending (wages plus transfer-fee amortisation), season by season, and average.

In 2016/17 City won 78 points and spent £356M on players: 0.22 points per £1M. Over the three seasons of the PSR window the average is 0.25. Over all nine breach seasons, 0.27. Adjust the earlier seasons for the growth in football spending, as Everton successfully argued should be done, and it falls to about 0.21.

Call it 2.1 to 2.7 points for every £10M.

City’s methods

City’s economists would say a club’s own average rate is the wrong measure for a club at the top. Their argument is diminishing returns: the first £100M turns a relegation side into a mid-table one, but the last £100M at a club already winning most of its matches buys much less. They have three ways of putting a number on that.

• The league-wide rate, across every Premier League club: 1.5 points per £10M in 2016/17.

• The league rate at City’s level of spending, where each extra pound buys less: 0.7.

• A match-by-match model of relative spending, the approach Everton’s expert took: also 0.7.

Same club, same season, and the exchange rate differs by a factor of four depending on who is doing the sum.

In Burnley v Everton this question did not need a clean answer, because Everton was a mid-table spender. The Commission accepted that diminishing returns exist in general and said they bite at clubs spending much more than Everton. City is the club that discussion was about. So this is one of the points where the precedent helps City more than Arsenal.

Step 4: Turn The Points Into a Top-Four Place

This is where I expected to copy and paste the Burnley method. It turned out to be the most interesting part of the whole exercise, because the method does not survive the move.

Here is what Burnley’s experts did, in one sentence:

• weaken Everton,

• replay Everton’s 38 matches 100,000 times,

• leave every other match as it happened, and

• count how often each club is relegated.

I did the same for City in 2016/17. I built ratings for all twenty clubs from the betting market’s closing odds, as in Part 2, weakened City, replayed City’s 38 matches 100,000 times and counted how often Arsenal finish in the top four.

Then I ran it with nothing taken off City at all, as a check.

Arsenal made the top four in 81% of those seasons.

That cannot be right as a description of what happened. With nothing removed, we know what happened: Arsenal finished fifth.

So where does the 81% come from?

Why the Burnley test breaks

Replaying City’s matches does not only reshuffle City’s points. It reshuffles the points of everyone City played.

In the real 2016/17, Liverpool took four points from their two matches against City: a 1-0 win and a 1-1 draw. Arsenal took one: a 2-1 defeat and a 2-2 draw. That three-point swing is the entire gap between fourth and fifth. Replay those four matches and, on average, the two clubs do about equally well against City, and Arsenal’s actual lead over Liverpool in all the other matches puts them fourth most of the time.

In other words, the replay hands Arsenal better luck against City, and calls it a result.

Burnley’s case had the same feature, but it did not matter much there. It was two clubs fighting over one place, and the experts compared the two directly: who goes down more often, Everton or Burnley?

Here there are three clubs, City, Liverpool and Arsenal, fighting over two places. “Does Arsenal make the top four?” can come out yes without City’s breach having anything to do with it.

City’s economists would spot this in the first hour. So I built two stricter tests.

Three tests

• Test A: replay City’s season, and ask whether Arsenal make the top four. The Burnley method as it stands. It says 81% before a pound is removed, so it cannot tell the breach from the dice. I would not rely on it, and I do not think a tribunal would.

• Test B: replay City’s season, and ask whether Arsenal finish above City. This is the head-to-head question, the nearest thing to “Everton or Burnley?”. With nothing removed, Arsenal finish above City in 41% of replays. It passes 50% once City are 1.7 points weaker.

• Test C: same luck. This one keeps the season as it actually happened and changes only City’s strength. Think of each City match as having had its own roll of the dice. City beat Arsenal 2-1 in December 2016. A slightly weaker City, with the same roll, might still have won. A much weaker City, with the same roll, would have drawn or lost. In this test a weaker City can only do the same or worse in each match than it really did; and when City drop points, their opponents that day pick them up. With nothing removed, this test gives back the real league table exactly: Arsenal fifth, 0%. That is the property the Burnley method lacks. And it passes 50% once City are 3.3 points weaker.

Why 3.3 and not exactly three, the gap in the table? Because City also had the better goal difference, so City needed to fall four points behind their real total, or Arsenal needed to gain from their own matches against a weaker City. Sometimes three and a half points of weakness produces that, sometimes it does not. At 3.3, it happens just over half the time.

• What Arsenal would argue: test B. It is the comparison the Commission itself relied on.

• What City would argue: test C, or something stricter still.

I think test C is the more defensible of the two, because it is the only one that starts from what actually happened. So from here on I hold Arsenal to it.

How much money is that?

Now we can join steps 3 and 4. How much money has to come off City for City to lose 3.3 points?

• With City’s own rate, the method the Commission accepted: £12M to £16M.

• With the league-wide rate: £23M.

• With City’s best rates, the league rate at the top or the match model: £44M to £50M.

Notice what this does. It turns the argument around. We no longer need to know exactly how much money is in play. We only need to know whether it is more than about £50M. If it is, Arsenal pass the strictest test under every exchange rate, including the ones most favourable to City.

Who wins?

Put the five ways of counting the money from step 1 against the seven exchange rates from step 3. That is 35 combinations. To be cautious I use the lower of my two assumptions about the money, the even spread.

Arsenal’s claim passes in 25 of the 35.

Look at where City win. The whole bottom row: the excess over the limit, spread across three seasons, £7M on these assumptions. And the right-hand end of the row above it: the excess after deductions, £22M, combined with City’s preferred exchange rates.

The top three rows are red all the way across. Once the money counted against 2016/17 is £92M or more, the exchange rate stops mattering.

So this is what I take from the simulation. The fight over the exchange rate, which looked decisive in Part 2, is the second most important fight. The first is how much money is attributed to the season. City need to win that argument; winning the exchange rate alone is not enough.

One more thing a careful reader will ask. In Burnley’s case the margin was wafer-thin, 50.5% against 47.6%, and Burnley still recovered its whole loss. The same would apply here: if a tribunal accepts causation on the balance of probabilities, Arsenal recover the full loss, not a percentage of it. City would argue for a percentage, the “loss of a chance”. The Burnley decision did not go that way, though it also never discussed the alternative.

Step 5: Price The Season

Suppose Arsenal clear causation. What did fifth place cost?

The cleanest evidence is in Arsenal’s own accounts. 2016/17 was a Champions League season; 2017/18 was a Europa League season. Both fell under the same Premier League TV deal.

Revenue fell by £29.9M: £18.5M in broadcasting, £10.3M in commercial income and £1.1M on matchdays. Arsenal’s own results statement that year put the fall mainly down to playing in the Europa League rather than the Champions League.

That is the starting point. From there, each side would pull in the direction you would expect, and we know from Part 2 how wide that gap can get. Burnley’s accountant said £51.7M. Everton’s said Burnley had made a profit.

What Arsenal would argue

The full £29.9M. Plus the prize money for finishing fourth instead of fifth, about £1.9M. Plus the knock-on: Arsenal did not return to the Champions League until 2023/24, and Champions League money buys players who help you qualify again. Burnley’s accountant handled that kind of uncertainty by building four paths and weighting them equally, and the Commission accepted it. The equivalent here is the Champions League lost for one, two, three or four seasons, 25% each.

What City would argue

Only broadcasting can be tied to the Champions League; commercial income moves for many reasons. Missing the Champions League also saves money, because players’ contracts carry bonuses for it. And one season only: in 2017/18 Arsenal finished sixth, twelve points off fourth, and that was nobody’s fault but their own.

I think City would win the argument about knock-on seasons. The Commission cut Burnley’s most optimistic assumption (finishing 13th every year became 15th), and refused to look beyond the years for which real figures existed. A tribunal is unlikely to hand Arsenal four seasons of Champions League money on the strength of one missed place.

Step 6: Add Interest

The Commission used Burnley’s own average cost of borrowing, compounded annually: 7.6% rising to 11.8%.

Arsenal borrow far more cheaply. From the accounts, interest paid as a share of borrowings was about 6.5% in 2017/18, fell to 2% to 3% after the stadium bonds were replaced by a loan from the owner in 2020, and has been around 5% since.

Compounded from May 2018 to today, that multiplies the loss by 1.43. So interest adds 43%, against 35% for Burnley over four years at much higher rates. Eight years of waiting does that.

The claim, assembled

Four versions.

• Arsenal’s ask: about £98M. The full revenue gap, four equally weighted paths, plus interest.

• A Commission-style middle: about £37M. One season, with 80% of the lost revenue reaching profit (the 80% is my placeholder for saved bonuses and costs; I have no data on Arsenal’s contracts), plus the prize money, plus interest.

• City’s fallback: about £21M. One season, broadcasting only.

• City’s main case: nothing. No causation, no loss.

If I had to pick the most likely landing zone, should the claim succeed, it is the middle one. And £37M is almost exactly what Burnley were awarded. That is a coincidence, but it is a useful one: it tells you the scale. A single missed Champions League place is worth about the same as a relegation was to Burnley.

It also tells you something about the bigger picture. The Commission found more than £830M of overstated income. The best single claim Arsenal can build from it is worth a few tens of millions. For City, the danger was never one claim. It is ten clubs each bringing two or three.

Where City would fight

Put the whole thing together and City’s defence has four lines, in the order I think they matter.

1. The amount. Only the excess over the spending limit counts, after every deduction, spread across the assessment period. On my assumptions this is the only argument that wins the case on its own.

2. The test. The Burnley simulation gives Arsenal 81% with no breach at all, so it proves nothing; apply a strict test. This costs Arsenal some ground, 3.3 points needed instead of 1.7, but does not defeat the claim unless the amount is small.

3. The exchange rate. Each extra pound buys less at the top. This cuts the points by a factor of up to four. Again, it only decides the case if the amount is small.

4. The damages. One season, broadcasting only, no knock-on. This is where City is most likely to win something.

And two that sit outside the economics. Whether a claim about May 2017 is still in time, which turns on when the clock started and on the Commission’s findings of concealment. And whether any of this survives City’s appeal.

The part worth keeping

I started this expecting the exchange rate to decide everything, because it decided so much of the argument between Burnley and Everton.

It does not. What decides Arsenal’s claim is a number in an unpublished appendix: how much of City’s disguised funding counts against 2016/17. If it is above roughly £50M, Arsenal clear the strictest test I could build, under every exchange rate City could argue for. If it is nearer £20M, City have a real defence.

The other thing I will keep is the 81%. A method that a tribunal accepted in one case gave a nonsense answer when I moved it to the next one, for a reason that has nothing to do with football finance: two clubs for one place became three clubs for two. Precedents carry reasoning, not software. Anyone running the Burnley playbook for another club will have to check that it still measures what they think it measures.

Boom — that was the damages claim Arsenal could bring against City.

Thank you for reading until the end ❤️

In Part 4, Liverpool: three titles lost to City by two points or fewer, only one of them inside the period of the Commission's financial findings. That is a very different claim, and in two of the three cases a harder one.

See you next week,

Martin

PS. If you enjoyed this piece, you might also enjoy how Kevin De Bruyne used data scientists to negotiate his City contract and how City used a data science technique to find De Bruyne’s replacement.

P.P.S. If a tailor-made report (such as the one I did on Kai Havertz’s contract negotiation) is what interests you, drop me a line @ martin.angelov@datadealsfc.com

The City Files is a series on the Manchester City verdict and the compensation claims that could follow it.

• Part 1: Manchester City’s 115 Charges, Explained in Plain English

• Part 2: The £35M Playbook: How Burnley Made Everton Pay, and How City’s Rivals Could Do the Same (you are here)

• Part 3: What Arsenal Could Claim

• Part 4: What Liverpool Could Claim

• Part 5: What Manchester United Could Claim

Hi friends,

In Part 1 of the series, we went through Manchester City’s 115 charges and what the independent commission ultimately found.

Today, I’ll look at what could happen next.

Within hours of the verdict, the BBC reported that several Premier League clubs had sought legal advice on whether they could claim compensation from City.

And this is not starting from scratch. There is already a precedent — and a very recent one.

In June 2026, a Premier League commission ordered Everton to pay Burnley £35M after finding that Everton’s breach of the financial rules had helped keep them in the Premier League in 2022 at Burnley’s expense.

To date, it is the only decided case that shows, in detail,

how a breach of football’s financial rules can be turned into causation, lost points and ultimately a damages figure.

So anyone thinking about bringing a claim against City is going to study it very closely.

That’s what I’m going to do today:

• Break down the economics of the case,

• Look at the arguments on both sides, and

• Reconstruct the Commission’s calculations step by step.

Let’s get into it.

A little bit of context first

On 22 May 2022, the last day of the season, Burnley lost 2-1 at home to Newcastle and were relegated. They finished 18th, 4 points behind Everton, with a better goal difference.

Ten months later the Premier League charged Everton with breaking its Profitability and Sustainability Rules (PSR) in that same season: Everton had lost £19.5M more than the rules allowed. A commission deducted 10 points, cut to six on appeal.

Burnley’s argument was simple to state:

Without the breach, Everton would not have had the squad that kept them up, and Burnley would have survived instead. So Everton should pay for what relegation cost Burnley.

Burnley’s compensation claim was brought under the Premier League’s own disciplinary rules and heard by the same commission that had dealt with Everton’s breach. It was put on hold until the breach case, including Everton’s appeal, was finished. The hearing took ten days in autumn 2025, and the decision is dated 2 June 2026, four years after the relegation it compensates.

Burnley did not have to prove Everton’s breach all over again (!). The league had already established the breach and its size. Burnley’s job was now to show that the breach harmed them — and put a number on that harm.

Other clubs had tried to claim compensation before, but those cases either settled or never produced a full decision. What makes Burnley v Everton different is that a tribunal actually decided the case, explained its reasoning and put a number on the loss.

And that is where the economics starts.

So how do we quantify the damage suffered?

It boils down to 6 steps.

Step 1: Start from the breach that has already been proved

The Commission treated the findings from Everton’s original case — including the £19.5M overspend — as binding in the compensation proceedings.

Burnley therefore did not need to prove the overspend again, and Everton could not argue that it was smaller.

But the principle worked both ways: when Burnley tried to rely on other alleged breaches that had never been established, the Commission refused. Compensation could only be based on breaches actually found.

That matters for City’s rivals too. Their starting point would be whatever breaches and amounts ultimately survive City’s own proceedings.

But knowing that a breach happened is not enough.

One needs to ask next, what would have happened if it had not?

Step 2: Build the “what if” world

Economists call this the counterfactual: the world in which Everton complied with the rules.

Everything that follows depends on it, because

Damages are essentially the difference between that world and what actually happened.

Everton argued for the least disruptive version. Without the breach, it said, it could simply have sold another player before the 30 June 2022 year-end, or avoided some earlier signings and sold players who were underperforming anyway.

The Commission rejected both. Its test was practical: what would a club acting in good faith, and in its own commercial interests, realistically have done?

Its answer was that Everton would have needed to raise another £19.5M from player sales earlier in the 2021/22 season.

In plain English:

Without the breach, Everton would probably have had a weaker squad.

That gets us halfway there. We now have:

• Everton overspent by £19.5M.

• Without that overspend, its squad would likely have been weaker.

And this begs the logical question of how much weaker points-wise would Everton have been? Or in other words,

How many points was that £19.5M worth?

Step 3: Turn the money into points

This is where the economic experts on both sides started disagreeing.

• Burnley’s experts estimated that Everton’s overspend had been worth roughly 3.7 to 7.1 points.

• Everton’s expert got as little as 0.2 to 2.6.

Same £19.5M. Completely different answers.

Why?

Because there is no observable exchange rate between pounds and Premier League points.

The experts had to build one. And they built it quite differently.

What Burnley’s experts did

Burnley’s experts started with player spending: wages plus transfer-fee amortisation. A £50M player on a five-year contract, for example, contributes £10M a year in amortisation, on top of his wages.

First, they checked the basic premise: do clubs that spend more actually win more points?

Across twelve Premier League seasons, the answer was clearly yes. The reported correlation between player spending and points was 0.72 (I rebuilt the same calculation from clubs’ published accounts and got 0.68 - not identical, but telling the same story).

Then came the important part.

They calculated Everton’s own points per £1M of player spending and averaged that rate over four different periods (see figure below). Depending on the window, multiplying that rate by the £19.5M overspend produced 3.85, 4.35, 4.82 or 7.13 points.

Everton’s published accounts contain everything needed to reproduce those numbers, so I did.

All four come out exactly.

But the chart also exposes the weakness of Burnley’s preferred 7.13-point estimate. In 2012/13, Everton won 0.86 points for every £1M it spent on players. By 2021/22, that had fallen to 0.17. Everton had not suddenly become five times worse at spending money. Football had simply become much more expensive while the number of points available remained the same.

Everton made exactly that argument.

Once earlier seasons were adjusted for the growth in Premier League spending, Burnley’s four estimates narrowed to roughly 3.7 to 4.7 points. The Commission accepted that Everton’s criticism had ‘some force’ and did not rely on the 7.13-point estimate. Even without the 7.13, every other estimate, adjusted or not, sat between 3.66 and 4.82 points: close to, or above, the four that mattered.

And that matters because Everton finished exactly 4 points above Burnley.

Everton’s expert had a very different answer

Everton’s expert, Derek Holt, asked a different question.

Burnley’s experts asked:

how many points has Everton historically won for each £1M it spent?

Holt asked:

when two clubs meet, how much does the gap in their spending change the result?

To answer it, he took every Premier League match in the four breach seasons and measured how much more the home side spent on players than the away side. Then he estimated how the chances of a home win, a draw and an away win shift as that gap grows. (The statistical tool is called an ordered logit; the gap is measured as a ratio of the two clubs’ spending.) With that in hand, he could take money away from Everton, replay its matches, and count the points it would have lost.

The decision describes the model well enough to rebuild it, so I did, from the same club accounts and every result from 2018/19 to 2021/22. The easiest way to see what it does is one match.

Put Everton at home to a club with exactly the same wage bill. The model gives Everton a 42% chance of winning, worth about 1.52 points on average. Now take £19.5M off Everton. It spent about £230M on players that season, so that is a cut of under 9%, and the chance of winning drops to about 40%: roughly 1.45 points. The difference is 0.07 points a match. Over 38 matches it adds up to about 2.5 points.

That is the first reason his number is smaller. His model is built on spending relative to the rest of the league, and in it £1M is worth about 0.13 points. Everton’s own record over the breach seasons said 0.20.

The second reason is how much of the £19.5M he counted against 2021/22. The assessment period covered four seasons but, because of Covid, the rules averaged 2019/20 and 2020/21 together, so it had three parts. Holt’s preferred version spread the overspend evenly, £6.5M for each part, which leaves only a third of it in 2021/22. Spreading it in proportion to Everton’s losses leaves just £1.6M. Only when he put all £19.5M into 2021/22, a version he himself called unrealistic, did he reach 2.6 points.

My rebuild gives 2.52, 0.82 and 0.20 points for those three versions, against his published 2.6, 0.8 and 0.2. It also picks the right result in 55.3% of 2021/22 matches, against the 55.0% he reported and about 59% for the bookmakers.

There is one more thing the rebuild shows. Fed only spending, the model expected Everton to win about 54 points in 2021/22. They won 39. That was the Commission’s criticism: before any adjustment, his model’s league table was simply the clubs ranked by what they spent.

So the disagreement was not really about arithmetic. Both sides’ numbers can be rebuilt from public accounts almost to the decimal. It was about

which model should turn money into points, and how much of the money belonged to the season that mattered.

The Commission preferred Burnley’s approach. Among other things, it found Holt’s model less compelling because it reduced football almost entirely to spending and predicted match results less accurately than the bookmakers.

Which exchange rate?

This is the first question that becomes especially important for Manchester City. The exercise is, in effect, about choosing an exchange rate between pounds and points.

• The club’s own rate. This is essentially Burnley’s approach: use the offending club’s own historical points per £1M.

• The league-wide rate. Estimate how much an additional £1M is associated with across all Premier League clubs.

• The rate at the top. If returns diminish as spending rises, another £1M may buy much less for an already dominant club than for a mid-table one.

Burnley v Everton does not settle that question for a club like City. The diminishing-returns discussion in the case specifically recognised that the position could be different for the biggest spenders.

I’ll come back to that when we run the numbers for individual City rivals.

For now, the Commission had an estimate of how many points Everton’s overspend may have bought. But four fewer points does not automatically mean Burnley stay up.

For that, they had to replay the season.

Step 4: Turn points into relegation

The experts took pre-season bookmaker ratings for all twenty Premier League clubs, weakened Everton by the estimated effect of the overspend, and simulated Everton’s 38 matches 100,000 times.

Every match not involving Everton stayed exactly as it happened.

Then they asked a simple question:

Who ends up in the bottom three more often: Everton or Burnley?

In every scenario, the answer was Everton. Even in the most cautious version, Everton were relegated in 50.51% of the simulations and Burnley in 47.59%.

That difference looks tiny. But legally, it was enough.

The Commission was applying the civil standard of the balance of probabilities. It did not need Burnley to prove with certainty that Everton would have gone down.

It needed to decide whether, without the advantage created by the breach, it was more likely than not that Everton rather than Burnley would have occupied the relegation place.

Every scenario pointed the same way.

I rebuilt this too

The experts’ Spreadex ratings are private, so I could not reproduce them directly. Instead, I used Pinnacle’s closing odds for every Premier League match in 2021/22 to build my own market-based ratings. Then, for each of Burnley's experts' eight estimates (3.66 to 7.13 points), I weakened Everton by that many points and replayed its season 100,000 times.

Across all eight scenarios, my relegation probabilities come within 0.7 percentage points of the published figures. In the most cautious scenario, I get Everton at 50.3%, against the experts’ 50.51%. Burnley comes out at 47.8%, against 47.59%.

So the core method behind the decision can be reconstructed almost entirely from public data.

And once the Commission accepted that Everton’s breach had caused Burnley’s relegation, the question changed again.

It was no longer: did Burnley lose money?

It was: how much?

Step 5: Turn relegation into £35M

This is where the accountants took over.

Burnley’s loss was measured as the difference between what actually happened after relegation and what the club would probably have earned had it stayed in the Premier League.

But there is an obvious problem: nobody knows what Burnley’s next four seasons would have looked like if it had survived in 2022.

So Burnley’s accountant built four alternative paths.

In every one, Burnley stays up in 2021/22. After that, the assumptions change: in some paths it remains in the Premier League; in others it is relegated later. Each scenario received an equal 25% weight.

And here the experts disagreed even more dramatically than they had over points.

• Burnley’s accountant calculated a loss of £51.7M.

• Everton’s accountant said Burnley had lost nothing. On his main case, Burnley was actually £18.2M better off because relegation reduced costs and generated player-sale income.

The Commission landed between the two.

One important adjustment went Everton’s way: Burnley had assumed it would finish 13th in the Premier League; the Commission considered that too optimistic and used 15th instead.

After working through the different assumptions, it arrived at:

• £24.6M of lost operating profit

• £1.4M of lost player-trading income

That gave a loss of £26M.

Step 6. Add interest

Then came something that could become particularly important for City’s older seasons: interest.

The Commission used Burnley’s average borrowing costs and compounded them annually. That added another £9.1M.

And that is how a £26M of loss became a £35M award.

What the precedent does, and does not, tell us about City

Put the steps together and a claim against City would have to do six things:

1. rest on breaches the panel actually found;

2. build a realistic “what if” world;

3. turn the money into points;

4. show, through a simulation, that without City’s advantage the claimant was more likely than City to finish on the right side of the line that mattered;

5. price the seasons that followed; and

6. add interest.

1. Established breaches

The published core decision changes the first step substantially. The Commission found that City’s income was overstated by more than £830M between 2009/10 and 2017/18, through what it calls a Disguised Funding Scheme, and that City breached both the Premier League’s and UEFA’s spending limits “by a very substantial amount” in every season charged, even on City’s fallback case that the sponsorship deals should simply be revalued at a fair market price. That is a far larger breach than Everton’s £19.5M. The season-by-season figures are in appendices that have not yet been published, and City have until 2 October to appeal. Whether the findings would bind later compensation claims in the same way as the Everton findings bound Burnley’s would depend on the rules that apply and on the outcome of any appeal; the structure of the Premier League’s rules suggests they would, but that is for lawyers to confirm.

What Burnley v Everton gives claimants is a tested method and a set of reasoned answers to the obvious objections. What it does not give them is a guarantee. It is a decision of one commission on one club’s facts, not binding precedent, and several of its findings would be argued afresh in a City case:

2. The “what if”

Everton could not show a realistic way to raise £19.5M without weakening the team. City would argue that a club of its commercial pull could have raised similar sums legitimately, from genuine sponsors or owner funding within the limits. The counter-argument is the Commission’s own finding: the deals were shams precisely because the sponsors were not paying the recorded fees.

3. The exchange rate

The Commission’s finding on diminishing returns was about Everton, a mid-table spender. It did not decide what rate applies to a club at the very top, and City would argue that an extra pound bought very little there. Claimants would reply that the Commission accepted the benefit of overspending is cumulative, and that City’s alleged funding ran across nine seasons. The cumulative finding was made on Everton’s facts; whether it carries over is part of the argument.

4. The places at stake

Burnley lost Premier League status. City’s rivals mostly lost titles, worth a few million pounds in prize money, or Champions League places, worth tens of millions a season.

As for the comparison, in Burnley’s case the question came down to whether Everton, without the advantage, were more likely than Burnley to be the club relegated. A rival of City would need to show something similar for the place it lost. And because Burnley recovered its full loss once causation was established, the same approach would apply in principle, although a different tribunal could be asked to take a different view on the facts.

5. Time & Interest

In English law, claims for breach of contract generally have to be brought within six years, but where a breach has been deliberately concealed the period does not start until the victim discovered it, or could reasonably have done so. Those are established rules. Whether they apply to compensation claims under the Premier League’s rules, and if so when time started running (Der Spiegel’s reports in 2018, the charges in 2023 or the Commission’s decision), are open questions, and could be the first battle any claim faces. The Commission’s findings that City concealed the scheme from its auditors and regulators would be central to that argument. Burnley v Everton did not deal with time limits at all.

And finally, a claim about seasons from a decade ago would carry a decade of interest, on whatever basis a tribunal chose.

Boom — that was the precedent all economic consultants will likely rely on to build their follow-on damages cases against Manchester City.

Thank you for reading until the end ❤️

I hope this makes the claims to come easier to follow.

In Part 3, I’ll run the same playbook for Arsenal: what their economists could argue, where the numbers take us, and how City’s side would likely push back.

Subscribe to not miss it.

Subscribe now

See you next week,

Martin

P.S. This post was AI-assisted. The model wanted a share of the £35M. I told it to prove causation first. It couldn’t.

She was 28, had just over a year left on her deal, and wanted two things: to stay at City and get a significant salary increase.

A little bit of context if you, like me, just started following women’s football.

Greenwood was one of the best defenders in the Women’s Super League (WSL), which is the highest tier in English football. She was also an England international. Pretty much by every account, an elite player. And yet, she was still facing a version of:

“We don’t pay women that much.”

Fast forward to December 2022 when she signed her new three-year contract with on significantly more lucrative terms.

How did she do it?

She brought data scientists to the negotiation table.

The same team that helped Kevin De Bruyne negotiate his €80M contract extension. The consultants had a pretty unusual problem to solve though. City already knew Greenwood was good. The difficult part was putting a price on how good she was.

Think about it. In the men’s game, you have transfer fees. Public salary estimates. Hundreds of comparable deals. If you want to argue that your client deserves a £60k-a-week raise on her current £120k, there are plenty of players you can point to.

Women’s football in 2022 was different. Hardly any transfer fees. No public salary database. Very few obvious benchmarks.

So how exactly are you supposed to benchmark that?

Well, Harvard Business School (HBS) did a case study on the topic and today I’ll deconstruct what (likely) happened behind closed doors.

So here are the 7 inventive methods her data scientists used to get her the raise she wanted.

1. Nobody moved the ball like Greenwood

I know I just promised you inventive methods, but before we get to them, let’s establish her playing time. Since joining City during the 2020/21 season and until the cut off date of 1 December 2022, Greenwood was an established starter that played 46 of the available 50 league matches and 88% of the available minutes.

I’m not going to make a big case out of his reliability (in the way Ben White did), but just consider that in the women’s game in England during that same period, the median outfield player played 47% of all available minutes.

Ok, now to the fun part. The first advanced metric used in her favor, according to the HBS case study, was illustrating what she did with ball. Analytics FC used their proprietary goal difference added metric. Now, obviously I do not have that, so a built a simpler expected threat model from the publicly available Statsbomb dataset. I won’t bore you with the maths. Basically, it asks a simple question: how much closer to scoring did her actions move City?

Well it turns out nobody moved the ball into danger like Alex Greenwood.

On that measure alone, during the 2020/21 season she added 0.42 xG per 90 minutes from ball progression.

What’s more, she ranked 4th of all 213 WSL outfield players on ball-progression value per 90 (attackers included).

And if you are like me and enjoy poking into other people’s arguments, yes she did take corners and freekicks, (which explains half of the xG), but even on open play alone she sits at the 82nd percentile of forwards, which is still better than four in five forwards.

Okay, that’s a great start, isn’t it?

2. The harder the pass, the better Greenwood looked

The second killer argument was the passes she completed. Her data science team successfully illustrated that she completed harder passes than she was supposed to. Here’s what I mean her data team meant.

Instead of looking at plain pass completion rate, they looked into her expected passes. Yes, that’s the same ideas as xG but for passes. Instead of asking how many passes Greenwood completed, the model asks how difficult those passes actually were. I trained a model on 296,000 WSL passes. Based on where a pass starts, where it goes and the situation around it, the model estimates how likely an average player was to complete it.

Well, Alex Greenwood completed 83.7% of those when she should have done 79.1%, or 4.6 percentage points above expectation.

That’s an over-performance that puts her 2nd among the 179 WSL players.

If I use all available data (USA’s national women’s soccer league, 2019 World Cup and Euro 2022), she “falls” to 6th. Combine this with the fact that Greenwood attempted more passes and created more danger than any defender in the WSL (see first section), and you had a very clear picture. So she wasn’t just completing lots of easy passes. She attempted a lot, tried difficult ones, and still completed more than expected.

And then comes City’s counter-argument.

3. Yes, but a defender is supposed to defend

There was one obvious argument that was supposed to eventually surface. “While exceptional at creating threat, Greenwood was still a defender. So she is supposed to defend, isn’t she?” - would the Citizens’ negotiators say.

And her defending numbers did not look that strong. At least not elite level strong. Tackles and interceptions put her around the 29th percentile of WSL centre backs. If I was sitting on City's side of the table, I know I would certainly bring that up.

Instead of pretending those numbers were wrong, Analytics FC, turned the question around. Okay, but what does Man City actually ask its centre backs to do? Because you see, if your team has a lot of the ball, then maybe most of the time you would be expected to play with it.

The figure below presents the teams’ passes per defensive action of all defenders. City’s women made 10.7 passes for every tackle, interception, clearance or block. First in the WSL. City’s men made 15.8. First of 98 clubs across Europe’s big five leagues.

City have the ball more, so of course their defenders defend less.

Also, when compared to men’s team superstar defenders, Greenwood sit comfortably well among Aymeric Laporte, Ruben Dias and John Stones. On tackles plus interceptions, Greenwood stood at 29th percentile vs Stones and Dias’ 7th percentile and Laporte’s 2nd. Ouch. Are they bad defenders as well? I guess Pep didn’t know any better.

So no, she was not a bad defender. That’s what a City center back looks like. In fact, the more of the ball a team has, the less its center backs defend.

In 2020/21, City had 66% of the ball. Across WSL centre backs, every extra 10 percentage points of possession is associated with roughly 1.5 fewer defensive actions per 90. Against all WSL centre-backs she looks average: 45th percentile. Adjust for how much possession City have and suddenly she’s 86th.

Great, so what have we established so far? She is an ever-present exceptional chance-creating defender that defends well. So what is she actually worth to City?

4. What’s Greenwood’s value to City?

According to the HBS case study, Analytics FC simulated City’s season with and without Greenwood in an attempt to quantify her value to the team’s chances of winning the league. So that’s what I did as well. I rebuilt the 2020/21 WSL season using expected goals and replayed it 50,000 times. Yes, 50,000. (That’s literally one line in R if you know how to code, or know how to politely ask Claude Code.)

With Greenwood, City win the title in about 40% of my simulations. Replace her with the cover (Dahlkemper, Bonner, Morgan) they actually had and it falls to 34%.

Now what if she joins a rival? Put her at Chelsea and City falls further, while Chelsea moves the other way. That makes sense.

5. Could Greenwood actually leave?

Simulations are all fun and games. But was the threat of her joining a rival real? Well, let’s see which teams actually used their center backs in the same way City used theirs. It turns out that Chelsea and Arsenal were indeed the closest fits, and by some distance.

So her outside options were real. I don’t know if Arsenal or Chelsea actually wanted her. That’s not really the point. The point is that

Greenwood had realistic places to go if City said no.

Which brings us to the central question.

6. How do you price a woman footballer?

Well, you bring Virgil van Dijk. Obviously. (True story as per the HBS case study).

At the time, data on women football was barely existent. I mean it still is non-existznt today, what’s left four years ago. So considering the lack of any meaningful public wages or fees in the women football market, her consultants turned to the men’s game.

They compared her with elite male centre backs by asking how far each player stood above the average defender in his or her own competition.

On my simple on-ball index, Greenwood stood at 1.59 standard deviations (SD) above the average WSL center back. For comparison, Van Dijk (which Analytics FC used), stood at 0.97 above the average Premier league defender (2019/20) - we’re talking world-class title-winning Van Dijk.

You don’t need to be a statistician here. Greenwood stood miles above the average defender in her league. Van Dijk did the same in his.

And yet, when you show the few scarce wage data points, you see the discrepancy. In 2021/22, estimated annual wages were roughly £11.4M for van Dijk.

Nobody was asking City to pay Greenwood £11M a year. Obviously.

But get this. City Women’s turnover was £4.1M and the league’s wage cap allowed 40% of turnover to go to player pay (that’s £1.6M). Or in other words,

Around three in ten defenders in the men’s big five leagues earned more on their own than City were allowed to spend on the entire women’s squad (£1.6M).

So the van Dijk comparison was never ‘pay me van Dijk money’. It was:

If this level of relative excellence is worth elite money in one of your teams, what is it worth in the other?

That’s much harder question to brush away with ‘we don’t pay women that much’.

Subscribe now

7. The women’s football was (slowly but steadily) growing

And there was one last argument: women’s football was growing fast.

The consultancy reportedly showed revenues growing 30% to 50% a year while wages lagged. And the big stuff had barely landed in those accounts yet: a new broadcast deal, England winning Euro 2022 at Wembley, record attendances.

The public accounts available at the time only went to 2020/21, but even those already showed the direction.

Combined WSL club revenue had risen from £8.5M in 2017/18 to £18.7M in 2020/21, roughly 30% a year. And that’s through a pandemic.

So a three-year flat salary had a problem. Maybe it looked fair in 2022. By 2025? Quite possibly not.

So, how much did she actually get?

Nobody published the exact number. All we know is that the new deal was “significantly more lucrative”.

So here’s my best back-of-the-envelope estimate.

City Women’s wage envelope was about £1.6M. Assume a star player takes 10-12% of that, as leading players typically do in a squad this size, and you get roughly £160K-200K a year. For a sanity check, Leah Williamson was reported at around £200K at Arsenal at the time, and England’s first-choice centre-back belongs in that bracket.

Work backwards from Greenwood’s current estimated pay (about £220K, on Football Manager’s numbers), and a 2022 deal in the region of £180K a year looks plausible.

That would imply a raise of roughly 50%, from perhaps £120K when she joined City.

Not bad. And she kept delivering.

She played 91% and 88% of City’s league minutes in the next two seasons, returned from a knee injury to start every game of England’s Euro 2025 win, and signed another City extension in 2025.

Today, Football Manager’s estimates put her among City’s best-paid players.

The defender.

Three years after: “we don’t pay women that much.”

I mean it’s still not that much compared to the men’s game. But then again the prize money and TV rights are not really comparable, aren’t they.

Boom — that was the Alex Greenwood story: how data helped her put a price on her value in a market that barely had one.

I hope you enjoyed this one as much as I enjoyed putting it together.

Thanks for reading until the end.

Talk soon,

Martin

PS. If you liked this piece chances are you’ll also enjoy how Kevin De Bruyne himself used data scientists to secure a 30% increase at 29, or how Man City used a magical data science technique to find his replacement or what Bruno Fernandes’ data science team likely argued in his ongoing negotiations with United.

P.P.S. If a tailor-made report (such as the one I did on Kai Havertz’s contract negotiation) is what interests you (including on women’s football), drop me a line.

The City Files is a series on the Man City verdict and the compensation claims that will follow it.

• Part 1: Manchester City’s 115 Charges, Explained in Plain English (you are here)

• Part 2: The £35M Playbook: How Burnley Made Everton Pay, and How City’s Rivals Could Do the Same

• Part 3: What Arsenal Could Claim

• Part 4: What Manchester United Could Claim

• Part 5: What Tottenham Could Claim

• Part 6: What Liverpool Could Claim

A note before you start: This piece was written before the Premier League published the Commission's decision on 29 September. What it confirms, and what it changes, is in an update at the end.

Hi friends,

Unless you live under a rock, you’ve probably already heard that Manchester City have been found guilty of virtually all the charges relating to their alleged breaches of the Premier League’s financial rules.

Over the next few posts, I’ll look at three things:

1. what City were actually accused of doing,

2. why they should be worried about what rival clubs do next, and

3. how much those clubs could realistically try to claim.

For the last two, I’ll use the methods used in the Burnley v Everton precedent as a reference point.

But first, the basics. You’ll hear the “115 charges” through everywhere on twitter, but very few people can say what the charges actually are. The short answer is that they are mostly about one question:

Did City’s accounts tell the truth about where the club’s money came from and where it went?

Everything else follows from that.

So in this piece I explain the allegations one at a time:

• what each one says,

• how the practice would work if it happened, and

• what it would change in the numbers.

For that last part I went through Manchester City’s own annual accounts, every year from 2003/04 to 2024/25.

One rule throughout. At the time of writing, the panel’s decision, and its reasons, remain unpublished. City deny wrongdoing and are expected to appeal. The money amounts attached to the allegations come from the Football Leaks documents published by Der Spiegel in 2018, and from the UEFA case that followed. City have always disputed the context of those documents. So everything below should be read as: if the allegations are as they were reported, this is what they mean.

Now that we got this out of the way, let’s explain how we got here.

1. How we got here

The case is much easier to follow once the dates are in order.

The story starts in September 2008, when Sheikh Mansour’s investment company, Abu Dhabi United Group (ADUG), bought the club.

In November 2018, Der Spiegel began publishing internal City emails obtained through Football Leaks, the platform set up by the Portuguese hacker Rui Pinto. They suggested that City had inflated the value of several commercial deals to get around UEFA’s financial rules.

UEFA opened an investigation in March 2019, and in February 2020 banned City from European competition for two seasons and fined them €30M, finding that the club had overstated its sponsorship income between 2012 and 2016. City appealed to the Court of Arbitration for Sport (CAS), which lifted the ban in July 2020. Most of UEFA’s allegations, CAS said, were either not proven or too old to be punished under UEFA’s five-year limit. It kept a €10M fine for City’s failure to cooperate.

The Premier League had opened its own investigation in 2019. In February 2023 it announced 115 charges. The hearing ran from September to December 2024, behind closed doors. Then came almost two years of silence, until The Athletic reported on 25 September 2026 that the independent panel had found City guilty on 114 of the 115.

In the nine seasons the charges cover, City won three league titles, two FA Cups and four League Cups.

2. The rules you need to know

Football’s financial rules rest on one simple idea: a club should not spend much more than it earns.

A worked example makes it concrete. Imagine a club that earns £300M a season from tickets, television and sponsors, and spends £350M on wages, transfers and running costs. It has lost £50M. Under the Premier League’s Profitability and Sustainability Rules (PSR), which took full effect in 2015/16, that is acceptable, as long as the club’s losses over three seasons stay under £105M and the owner covers them. UEFA’s Financial Fair Play rules, introduced from 2011, work on the same principle with tighter limits.

Now imagine the same club finds an extra £60M. What happens next depends entirely on where the money comes from.

• If a sponsor pays it, it is income. The club now earns £360M, has a £10M profit and can spend more next season.

• If the owner simply transfers it, it is not income. It covers the loss, but under the rules the club still lost £50M, and the owner cannot keep doing that beyond the limit.

The money is identical. The label decides whether it counts.

Four terms will come up repeatedly, so it is worth defining them now.

• Related party. A company or person closely connected to the club, such as its owner or a company the owner controls. Deals with related parties are allowed, but they must be disclosed and valued at a fair price. Otherwise an owner could inflate a club’s income simply by overpaying for a sponsorship (as defined above).

• Disguised equity funding. Owner money presented as something else, usually as commercial income from a third party. This is the central allegation against City.

• True and fair view. The standard that club accounts, and the financial information clubs give the league, must meet. The Premier League cannot audit every contract. It relies on clubs describing their finances accurately.

• Time-barred. Too old to be prosecuted under the applicable rules. This is why City escaped most of UEFA’s case in 2020. The Premier League’s rules have no such limit.

3. What the 115 charges are

When the Premier League announced its case in February 2023, the charges fell into five groups.

The structure matters more than the total.

• The 54 charges about financial information and the 14 about pay are the core: they say City’s accounts did not show where the money came from or what the club really spent.

• The 12 PSR and UEFA charges largely follow from them. If income was overstated and costs understated, a club can appear to comply with a loss limit it actually breached.

• The 35 cooperation charges concern the investigation itself.

That is also why the number is so large. A single practice is charged once for every season it affected and once for every rule it touched. As The Athletic put it this week,

The charges boil down to a handful of issues, multiplied by the years involved.

4. The core allegation: owner money presented as sponsorship

City’s principal sponsor is Etihad, the Abu Dhabi airline. City’s owner is Sheikh Mansour, through ADUG. Both are rooted in Abu Dhabi, which is exactly why the related-party question arises.

According to the leaked emails, much of the money City recorded as Etihad sponsorship did not come from Etihad. The clearest example is 2015/16.

City booked £67.5M of Etihad sponsorship that season. An email from City’s then chief financial officer, Jorge Chumillas, reportedly set out that

Only £8M of it was to come from the airline and the remaining £59.5M from ADUG.

It was not the only year. Here is everything I could trace to published reporting of the leaks and the UEFA case.

A second email points the same way. In December 2013, City board member Simon Pearce reportedly wrote to Etihad that, of the £99M the airline owed City for 2012/13 and 2013/14 combined (£31.5M and £67.5M), Etihad itself would provide just £8M. That puts the owner's share across those two seasons at about £91M, close to the £84M in the table.

Now go back to the £59.5M during 2015/16 recorded by City as sponsorship. Why would anyone route money this way?

• If the £59.5M is recorded as sponsorship, it counts as money the club earned. The club's income goes up by £59.5M, so it can spend £59.5M more on wages and transfers and still appear to break even.

• If the same £59.5M is recorded honestly, as money from the owner, it does not count as income at all. The club’s income stays where it was, and that extra spending shows up as a loss, and the owner can only cover that loss up to a limit: £105M over three seasons under the Premier League's rules, and far less under UEFA's, just €30M (about £22M) over three seasons by 2015/16. The £59.5M allegedly routed through Etihad that year was, on its own, roughly double UEFA's entire three-year allowance.

Same money, same bank account, same players signed. The only difference is the label, and the label decides whether the club looks like it is following the rules.

Here is where that money sat in City’s accounts.

City’s income comes from three sources:

• matchday (tickets and hospitality),

• broadcasting (television and prize money) and

• commercial (sponsors and merchandise).

Commercial is the one that changed fastest after the takeover. It was £25M in 2007/08 and £121M by 2011/12, the first year of the Etihad stadium and campus agreement. The red slices are the Etihad amounts the emails describe as owner money.

They are never most of City’s income. But they sit exactly in the category the rules use to decide how much a club may spend.

What happened when UEFA looked at this

In 2020, UEFA concluded that City had overstated their sponsorship income between 2012 and 2016. According to The Athletic, UEFA’s case put the total at around £200M: money City had recorded as sponsorship but which was, in UEFA’s view, really the owner’s, passed through sponsors. That is close to the roughly £180M in the table above.

That is not the end of the matter. UEFA’s case could only reach back five years, so much of it was thrown out as too old. The Premier League’s rules have no such limit, and its charges go back to 2009/10. Some of the evidence also came too late for CAS: an email from City board member Simon Pearce only became public two days after CAS published its full judgment. And according to The Athletic, the Premier League uncovered new lines of enquiry during the case. What the panel concluded about which amounts is not yet known.

Inside one season: 2012/13

The allegations become clearer when you look at a single year, both through the leaked emails and through City’s own accounts.

In the summer of 2013, as City’s finance team compiled the 2012/13 accounts, sacking manager Roberto Mancini had created a problem. According to Der Spiegel, Chumillas wrote internally that the cost meant City would be £9.9M short of complying with UEFA’s rules that season. The chief executive, Ferran Soriano, reportedly suggested the gap could be closed with a bonus sponsors had agreed to pay for winning the FA Cup.

Except, City had lost the FA Cup final to Wigan.

The emails then describe a workaround: payments from three Abu Dhabi entities (Etihad, the investment fund Aabar and the Abu Dhabi tourism authority) would be adjusted to cover the shortfall. When Chumillas asked whether they could change the dates of the sponsorship payments, Pearce reportedly replied:

“Of course. We can do what we want.”

For many people that line has become the whole case in six words. City have never disputed the contents of the emails, only the context in which they were presented.

City’s accounts for that year add something the emails do not. Look at the club’s “other operating income”, a line that is normally close to zero.

• In 2011/12 it was £12.8M, from selling intellectual property to related parties.

• In 2012/13 it was £47.7M: £22.5M from selling intellectual property to related parties, £24.5M to third parties and £0.8M of other income.

Selling intellectual property, such as the rights to use the club’s brand, to sister companies within the same ownership group is legal, and the accounts disclosed it. But it is a reminder of how much of City’s route to break-even in those years ran through companies connected to its owner. In the season the emails describe a £9.9M gap, the accounts show £22.5M of income from related parties that was not sponsorship at all.

The same accounts also show how the owner's funding was made permanent. After the takeover, City's losses were funded with loans from its parent company. By the end of 2011/12 the club owed its parent £822.9M, and its debts exceeded its assets by £468.8M. During 2012/13 it borrowed more, and then the whole balance, about £1.01bn, was converted into new shares. The owner stopped being a lender and became, in effect, a larger shareholder: the debt disappeared and the balance sheet swung from £469M in the red to £435M in the black.

Or in less technical terms:

• Before 2012/13: City owed its owner about £1bn and looked deep in the red.

• After 2012/13: City owed nothing, looked solidly in the black, and the owner simply owned more of the club.

This was legal, and it was disclosed. It just shows how much the owner put in, about £1bn by then and £1.39bn in total. But the charges aren’t about this money. They’re about the other money, allegedly passed through sponsors so it looked like income.

5. Allegation two: the manager’s second contract

The logic that works on income also works on costs.

If someone else pays part of a club’s expenses, the club looks cheaper to run than it is, and it can spend more elsewhere without breaching the limits.

Mancini managed City from 2009 to 2013 and won the club’s first league title in 44 years. His official City salary was reported at about £1.45M a year.

According to documents published by Der Spiegel, he also had a consultancy agreement with Al Jazira, a club in Abu Dhabi also owned by Sheikh Mansour, worth about £1.75M a year and requiring only a few days’ work. The reporting described a payment trail leading back to ADUG. If that is correct,

More than half of what City’s manager earned was paid outside City’s accounts.

In money terms it is the smallest allegation, about £7M over four years. In principle it is also the clearest, because a manager’s pay is a basic cost of running a football club, and the charges on player and manager pay say City did not disclose all of it.

6. Allegation three: image rights paid outside the club

The third allegation concerns image rights, which need a short explanation.

A footballer’s image rights are the right to use his name and face commercially: in advertising, on merchandise, in video games. Top players often hold these rights in a separate company, and their club pays that company to use them. Those payments are a legitimate part of what a player costs, and they belong in the club’s accounts alongside his salary.

The reporting on the charges describes a company called Fordham Sports paying part of certain City players’ image-rights fees between 2010/11 and 2015/16, around £11M a year, in an arrangement reportedly known internally as Project Longbow. If that is right,

Part of the real cost of City’s squad sat outside City’s books for six seasons: about £66M in total.

7. Putting it together: what the accounts would have shown

Each allegation pushes the numbers in the same direction.

• The sponsorship one makes income look larger.

• The pay ones make costs look smaller.

Together they make the club look more profitable, and therefore more compliant, than it was.

The first place this shows up is the wage-to-revenue ratio, the share of income a club spends on staff. It is the most widely watched warning sign in football finance.

The dark line is City as reported. In 2010/11 the club spent 114% of its turnover on wages: it paid out more in salaries than it earned in total. By 2015/16 the ratio was down to 50%, which on paper is

One of the fastest financial turnarounds any Premier League club has ever managed.

The red line applies the allegations: it removes the alleged owner money from income and adds the alleged off-book payments to wages. In most years the difference is modest. In 2012/13 the ratio moves from 86% to 101%, and in 2013/14 from 59% to 75%.

Those are also exactly the years UEFA was assessing. In May 2014 it fined City €60M (about £49M), with €40M suspended, limited their Champions League squad to 21 players and froze their wage bill for two years. The cost shows up in City’s 2013/14 accounts as a £16.3M settlement with UEFA.

The second place the pattern shows up is the bottom line.

The early years (09/10 to 11/12) barely change. City lost between £100M and £200M a year after the takeover, openly, and that was never the allegation. The years that change are the ones that looked like the turnaround. In 2015/16, a reported pre-tax profit of £20M becomes a loss of £51M.

The most telling comparison is over three seasons, because the Premier League’s rule is a three-year test. Over 2013/14 to 2015/16, the first window PSR assessed,

• City reported a combined profit of £7M.

• With the alleged items applied, the same three years show a loss of about £142M, against a permitted loss of £105M.

That comparison needs careful reading. The rules let clubs exclude certain spending before testing the limit, including youth development, women’s football and the depreciation of stadiums and training facilities. Those figures are not in the published accounts, so I have not applied them, and with them City’s adjusted loss could come back under £105M. This is not a PSR calculation and should not be quoted as one.

What it does show is how the charges connect. The PSR and UEFA charges are not separate accusations of overspending. They follow from the accounting ones:

A small reported profit becomes a large loss once the alleged items are put back where they belong.

8. Why the owner’s money is not the offence

Much of the commentary treats Sheikh Mansour’s investment as the scandal. Under the rules, it is not. An owner is allowed to put money into a club. The rules only limit how much of it can be used to cover losses.

City’s accounts show exactly how much the owner put in.

In the six seasons after the takeover, City lost £601M before tax. The owner covered it, first with loans and then, in 2012/13 and 2013/14, by converting £1.17bn of those loans into shares (the blue line that sky-rockets on the figure). In total the owner has put £1.39bn of equity into the club since 2008.

After 2014 the losses stop growing (the red line in the figure above). Apart from the Covid season, the club has roughly broken even for a decade and has not needed new equity since 2020/21.

All of that was disclosed, and none of it is charged. The allegation is narrower, and in a sense more serious, that

On top of the declared investment, more owner money arrived labelled as something else.

9. The cooperation charges

The last group, 35 charges, has nothing to do with money. It concerns how City responded to the investigation between December 2018 and February 2023.

The Premier League’s rules oblige clubs to cooperate with its inquiries: to provide documents, answer questions and act in good faith. The charges allege City did not.

There is a precedent. In 2020 CAS upheld the €10M fine against City for obstructing UEFA’s investigation, and was scathing about the club’s conduct even as it lifted the ban. And one of the leaked emails, from City’s head of legal in 2014, reportedly said

The chairman (Khaldoon Al Mubarak) would rather spend £30M on the 50 best lawyers in the world, suing for 10 years, than accept a sanction.

12 years later, the case is still ongoing.

These charges matter for two reasons.

• They carry their own sanctions, independent of the financial findings.

• And they help explain why the process took so long.

10. What happens next

Quite a lot, and slowly.

The sanction. The panel has found City guilty but has not yet decided the punishment. Under the Premier League’s rule W.51 it can impose anything from a reprimand or a fine to a points deduction or expulsion, and it can combine them. Most lawyers expect any points deduction to be applied going forward rather than by stripping past titles.

The appeal. City are expected to appeal, as they did successfully against UEFA in 2020.

The decision itself. The panel’s reasons have to be published, but both sides can argue over what is released and what is redacted. Until then, nobody outside the case knows which amounts the panel accepted, whether it found the same pattern in years the leaks did not cover, or which single charge failed.

The timetable. People close to the case have told The Athletic they would be surprised if all of this took less than another year.

And then the claims. Several Premier League clubs have reportedly instructed lawyers to assess whether they can claim compensation for places, prize money and Champions League income they lost to City. That is not a fantasy.

In June, a Premier League commission ordered Everton to pay Burnley £35M because Everton’s rule-breaking had kept them up at Burnley’s expense.

• How that case was won, step by step, is the subject of Part 2 of the City files coming up shortly.

• How other clubs such as Arsenal, Liverpool, Man United and Tottenham can claim their damages will be covered in Parts 3-6 of the series, including detailed break-downs of the likely amounts to be claimed.

To conclude

Three ideas make the whole case easier to understand.

1. The money is not the issue; the label is. The same £60M counts as income if a sponsor pays it and as owner funding if the owner does. Football’s financial rules depend entirely on that distinction, which is why the charges focus on whether City’s accounts described their money accurately.

2. The alleged items are small relative to City and large relative to the rules. A few tens of millions a year would not change how big the club was. On City’s own figures, it could change whether a reported profit was really a nine-figure loss.

3. “115” is a count of rules and seasons, not of schemes. Underneath it are a small number of alleged practices, one accounting consequence and one long dispute about cooperation. Once you see that structure, the rest of the case follows.

For an even more beginner-friendly review and a selection of the most interesting leaked documents follow the magic hat 🎩

Boom — that was my resume of Man City’s 115 Charges, explained in plain English.

In Part 2, we’ll move from what City did to what their rivals can do about it. We will take apart the 2026 Burnley v Everton case, the precedent that could prove most dangerous (money-wise) for City: the exact methods and reasoning a Premier League tribunal used to make one club pay another for breaking the financial rules, and that City's competitors can now use to seek follow-on damages of their own.

Thank you for reading until the end ❤️

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

Talk soon,

Martin

PS. If you enjoyed this piece, you might also enjoy how Kevin De Bruyne used data scientists to negotiate his City contract and how City used a data science technique to find De Bruyne’s replacement.

P.P.S. If a tailor-made report (such as the one I did on Kai Havertz’s contract negotiation) is what interests you, drop me a line.

This post was AI-assisted. The AI was paid entirely by me, and I can confirm none of it came from a related party.

11. Update, 29 September: what the decision actually says

A day after this piece was finished, the Premier League published a redacted version of the Commission’s core decision. It changes the picture in four important ways.

The charges are no longer allegations. The Commission found every financial charge proven. The one charge that failed, the “115th”, was one of the four about co-operating with the investigation. On the rest, it found that City had “made concerted efforts to stop and frustrate” the league’s inquiry. It also found that some of the club’s key witnesses gave evidence that was false in important respects, and that some of them knew it.

The scheme had a name, a start date and a method. The Commission calls it the Disguised Funding Scheme, devised in early 2010. The Abu Dhabi sponsor paid only part of each recorded fee, the “Base Sum”. The owner’s company paid the rest, the “Tagged Sum”, and City booked the whole amount as sponsorship income. The first use came at the very end of 2009/10, including contracts backdated to earlier in the season, and it had a very specific purpose: to stop City breaking Chelsea’s record for the biggest single-season loss in Premier League history, £140M. Documents were created to reassure sponsors they would never pay more than the Base Sum, and to hide the arrangement from regulators and auditors. The auditors, the Commission found, never knew.

The numbers are far bigger than anything the leaks could show. The amounts I could trace above to published reporting added up to about £180M, across four seasons and two sponsors. The Commission puts the overstatement of sponsorship income at more than £830M across all nine seasons, and the total effect, including costs kept off the books, at more than £900M. The season-by-season figures are in appendices the league has not yet published.

The image-rights deal was hiding in plain sight. In Part 1, I pointed to a £24.5M line in City’s 2012/13 accounts: intellectual property sold to “third parties”, which I described as legal and disclosed. The decision says the Fordham image-rights arrangement put £24.5M into City’s operating income that was in fact money from the owner, and kept £49.4M of image-rights costs out of the accounts. The year is redacted, but the amount and the description match that line almost exactly. One correction too: the Commission accepted that Project Longbow, the internal programme it sat in, was mostly a set of genuine attempts to raise revenue. Only the Fordham strand was a sham.

Two more findings matter for what comes next. Once the accounts are restated, City broke both the Premier League’s and UEFA’s spending limits “by a very substantial amount” in every season charged. And the conclusion holds even on City’s own fallback argument: revalue the sponsorship deals at a fair market price instead of treating them as shams, and City still breach the limits.

City have until Friday 2 October to appeal. The sanction will be decided at a separate private hearing.

It may come as a surprise to some of you, but one of the Premier League’s best full-backs (who also happens to be one of the most underpaid players in his position) is on the move, and for cheap. Yes, that’s Tyrick Mitchell I’m talking about.

And if you missed the memo, we’re talking about a chart-topping England international who is about to leave Crystal Palace for free in June (or in a few months if United and Liverpool pull the trigger early).

In today’s article, you’ll learn more about the Premier League’s most underpaid elite full-back.

Let’s get into it.

1. How good is Tyrick Mitchell ?

An elite Premier League full-back?

If I was on Mitchell’s negotiation team, I’d start with where he ranks among attacking full-backs. Over the last two seasons, Mitchell ranks 3rd and 4th among Premier League full-backs.

The index scores every Premier League full-back on six pillars, each a weighted set of Wyscout per-90 measures turned into a percentile against the rest of the league that season. For an attacking wing-back the weights are: attacking contribution 30% (xAG, chances created, accurate crosses and passes into the box, possession-adjusted), carrying and progression 25% (progressive runs, dribbles, passes into the final third), defending 25% (defensive duel win rate, sliding tackles, interceptions, recoveries), team goals conceded 10%, security on the ball 5% and aerial duels 5%. On 2025/26 Mitchell is at the 100th percentile for defending, 84th for attacking contribution and 77th for team defensive outcome, which is the 65% of the weighting; he is 24th on carrying, 37th on security, 26th aerially and 6th on build-up passing, which is the rest.

You can debate the construction of the index all you want, but one thing is sure: Mitchell was one of the (unsung) poster boys of Palace punching above their weight in recent seasons, and the data confirms it.

Now, speaking of his metrics brings us to the second point I’d lead the negotiation with.

Elite interceptions & duels won

Over those same two seasons, he has been among the best interceptors of the ball while also posting one of the highest duel win rates. That’s across 71 Premier League full-backs: he wins 68% of his duels (97th percentile) and makes on average around 5 interceptions per 90 (84th percentile). The two have nothing to do with each other across the league, which is exactly why hardly anyone is elite at both. He is. Widen the pool to Europe’s top 5 leagues, and that corner up to the right is still empty.

If we plot his expected assisted goals vs his defensive duels, you’ll see the only player close to him is Reece James. No need to explain much more.

Elite availability

Now another number that catches the eye: reliability. Over the 5 previous seasons, he’s played at least 80% of the club’s minutes. If this means nothing to you, just remember that only 2 full-backs managed that in the top 5 leagues. That’s him and Napoli’s Di Lorenzo.

2. The underpaid star

And yet, despite those impressive stats, he’s paid like a squad player. The figure below ranks the full-backs on my index (horizontal axis) and on their salary (vertical axis). It turns out he is at the top end of the performance spectrum and is paid below the median, but also well below what his level suggests.

Well, you could say full-backs are generally not the best-paid position, sure, but considering his level and his running-down contract, either Palace can keep an elite player without paying a transfer fee, or another Premier League club can get a bargain on an England international.

How big is the gap? Based on a wage model that predicts what a club of Palace’s size would pay a full back of his stature and age, Mitchell should be paid about 64,000£/week (more on the model in the next section). Being on 40,000£ currently, he stands at 37% below the benchmark. Of the 14 Premier League full-backs in the top tenth of the index, he is the furthest below his benchmark. Part of the explanation, of course, could be that he has not signed a new contract since 2022.

Okay, so what did we learn so far? He’s 27, an England international. Tops the rankings for two seasons straight (on our index). Tops the duels won x expected assists chart. And in 3 months could be gone for free (signature-wise). That’s a strong position from which to negotiate a salary bump, isn’t it? Let’s see his options.

Subscribe now

3. How much could Mitchell be paid?

If we try to predict a player's salary with a linear model (that we mentioned above), here's what he would be expected to earn at each club that has so far shown interest in him. The model regresses the log of the weekly wage on the position-index percentile, age and age squared, the log of the club's wage bill, the league and the position family. It is fitted on 1,449 outfield players in the top 5 leagues with 900+ league minutes in 2025/26 (Capology wage estimates), and it explains 70% of the variation in wages (R squared of 0.70).

He currently stands at £40,000/week, but the model suggests Palace would need to get him to at least £64,000. With the big boys, though, he could easily triple his wage to around £120,000 per week.

That’s of course according to the model. If we plot him against his potential future teammates, the model’s estimate could even prove somewhat conservative, since he wouldn’t even be the highest earner at United.

Is there a job to walk into?

Well, on the face of it, he looks like the perfect replacement once Luke Shaw is out of the club. Recent rumours suggest they won’t be extending him. So Mitchell looks like he could slot straight in. At Liverpool, he would compete with Kerkez, though on the face of it he might even be the starter.

4. So where should he go?

Okay, so where does that leave us?

Staying at Palace is not a bad option. Far from it. He knows the system, plays basically every minute and, if Palace really want to keep him, the data suggests his wage should move from around £40k to at least the £60-70k range.

However, my model values the same player at around £120k a week at Liverpool and a little above £100k at Manchester United. That is simply because the same level of player gets paid differently depending on the club he plays for.

So financially, the case for moving is pretty obvious.

The harder question is playing time.

Liverpool already have Kerkez, so Mitchell would be walking into proper competition. United looks a bit cleaner. If Luke Shaw does leave, there is a much more obvious slot to walk into, and Mitchell’s profile actually makes sense for a team looking for a reliable, high-volume left-sided defender.

So, if the choice really comes down to Palace, Liverpool or Manchester United?

SalaryData-wise, United probably makes the most sense.

Boom — that was my take on the Premier League’s most underpaid elite full back.

I hope you enjoyed reading it as much as I enjoyed creating it.

Thank you for reading until the end ❤️

Cheers,

Martin

PS. If you liked this piece chances are you’ll also enjoy how Kevin De Bruyne himself used data scientists to secure a 30% increase at 29, or how Man City used a magical data science technique to find his replacement or what Bruno Fernandes’ data science team likely argued in his ongoing negotiations with United.

A raise of around 25%.

It turns out White had hired the same data consultancy that famously helped Kevin De Bruyne prepare his Manchester City contract negotiation.

But White presented them with a very different problem.

De Bruyne could walk into a negotiation as one of the best players in world football.

White could not.

His performance numbers were good. Sometimes very good. But they did not scream best defender in Europe. So his analysts could not simply produce fifty pages of rankings showing that Arsenal were underpaying a superstar.

They needed to find something else.

And according to the Harvard Business School case study and ESPN’s account of the negotiation, they found it in something most football analysis treats almost as an afterthought:

Ben White was always available.

The challenge was turning that from a nice characteristic into something Arsenal should actually pay for.

So I tried to reconstruct the main analyses.

Everything below uses information available around the time of the negotiation. It goes without saying that I do not have the actual report, so this is a reconstruction of the logic rather than their exact numbers.

The result comes down to four analytical tools.

Before getting into the tools, it is worth understanding what made White’s negotiation interesting.

Why Ben White needed a different kind of argument

Most football statistics are measured per 90 minutes.

That makes sense if you want to answer:

How good is this player when he is on the pitch?

But Arsenal were not buying ninety minutes. They were committing tens of millions of pounds to a player over several years.

And for that decision there is another question:

How often is he actually available to produce those ninety minutes?

That distinction matters.

An exceptional player who is available for half the season does not necessarily deliver more value than a very good player who plays virtually every week.

White’s advisers appear to have built much of their case around that gap.

So the first tool was to measure it.

Tool #1: Measure the thing per-90 stats miss

How valuable is being there every week?

Start with the simplest evidence. Before signing his new contract, White had been extraordinarily available.

At Brighton in 2020/21, he played 91% of the club’s Premier League minutes.

At Arsenal: 84% in 2021/22, 91% in 2022/23 and roughly 86% of the available league minutes by the February 2024 cut-off used here.

Across his first 100 Premier League matches available to Arsenal, he appeared in 93. Transfermarkt records only 31 injury days across the preceding three and a half seasons.

That is a nice record. But Arsenal already knew Ben White rarely missed games.

The useful analytical question is:

How unusual was it?

So I compared him with every outfield player who spent all three complete seasons from 2020/21 through 2022/23 in the Premier League. That leaves 228 players.

White ranks 11th of 228. Among defenders, he ranks 3rd of 98. Only James Tarkowski and Tyrone Mings spent more of those three seasons on the pitch.

Suddenly ‘Ben White is reliable’ becomes something more useful:

Very few Premier League defenders were as consistently available as Ben White.

But the interesting part comes when we combine that with performance.

Good every week can beat great half the time

The HBS case describes White’s individual match statistics as solid rather than exceptional. My reconstruction reaches roughly the same conclusion.

Across ten role-adjusted full-back metrics, White sits around the 67th percentile during his Arsenal right-back seasons. Good. Not elite.

And that is important because the exercise should not manufacture an elite player where there isn’t one.

Instead, put two things on the same chart: quality per match and percentage of minutes played.

White is not at the very top vertically. He is almost all the way to the right.

Once I multiply his performance advantage over the median full-back by the share of minutes he actually played, White moves from roughly the 70th percentile per match to 13th of 126 full-backs in season-long contribution on this measure.

Put differently:

About seven in ten full-backs were worse per match. But roughly nine in ten delivered less total contribution over the season.

The comparison becomes even clearer if you turn the question around.

A full-back available for only 45% of the minutes would need to perform around the 89th percentile per match to deliver approximately the same season-long score.

That is the negotiation argument. White did not need to be one of the best full-backs in Europe every time he stepped onto the pitch.

He was very good. And he stepped onto the pitch almost every week.

What this gave White: a way to turn reliability from a compliment into measurable season-long value.

Subscribe now

Tool #2: De-risk the long contract

What happens when a 26-year-old full-back gets older?

Arsenal were not negotiating for the Ben White of March 2024. They were also negotiating for the Ben White of 2025, 2026, 2027 and potentially beyond.

That creates another problem.

Full-back is a physically demanding position. Pace, repeated carries and attacking involvement can become harder to sustain as players move into their thirties.

White’s report apparently addressed that directly.

According to ESPN, Analytics FC modelled the ageing curve of full-backs and considered what might happen if White eventually moved inside to centre-back. That mattered because White had already played there.

The argument was not:

Ben White will become an elite centre-back at 30.

It was more subtle.

Are the skills Arsenal are paying for the ones most likely to survive ageing?

I tried to test that.

For full-backs and centre-backs, I followed players from season to season and measured how different skill groups changed after age 26.

The pattern is useful.

• Carrying, progression and attacking contribution tend to decline more sharply.

• Build-up passing is much flatter.

• Ball security actually improves in this sample.

Now compare that with White.

• Across his Arsenal right-back seasons, his build-up passing was about 1.2 standard deviations above the average Premier League full-back.

• His ball security was roughly 0.9 above average.

Those were two of his strongest attributes. And they were also two of the skills that held up best with age in the sample.

His defending score was weaker, and moving inside would not magically fix that. So the argument should not be pushed further than the evidence allows.

But the basic point survives:

White’s value was concentrated in skills that looked relatively durable, while his ability to play centre-back created another possible use for him later in the contract.

That changes the risk of a long deal.

Arsenal were not simply buying four more years of the exact same right-back. They were buying a player whose strongest skills had plausible routes to remain useful as his physical profile changed.

What this gave White: evidence that the contract did not necessarily expire when his peak full-back years did.

Tool #3: Make the outside options credible

Who else could actually use Ben White?

A player negotiating with his current club needs another thing:

Alternatives.

Not necessarily a transfer offer sitting on the table. But enough evidence that walking away is plausible.

According to ESPN, Analytics FC simulated White in the starting elevens of five clubs: Manchester City, Barcelona, Milan, Bayern Munich and Marseille. The real analysis used Analytics FC’s proprietary Goal Difference Added model, which I cannot reproduce.

But one part of the exercise can be tested:

Would White actually fit how those teams played?

For every club in Europe’s five big leagues, I built a fifteen-metric profile of what its full-backs were asked to do: passing, progression, crossing, box involvement and defending.

Then I compared each club’s profile with White’s Arsenal right-back profile. The closer the profiles, the smaller the distance.

Arsenal come first — unsurprisingly, because White helped create Arsenal’s own profile.

But after that:

• Milan rank 3rd;

• Manchester City 4th;

• Barcelona 9th;

• Marseille 20th.

Among the clubs with wage bills above €150M, City and Barcelona are the two closest stylistic matches to White in Europe.

Then add squad composition.

• At City, Kyle Walker was 33.

• At Milan, the right side was shared between Davide Calabria and Alessandro Florenzi, who was also 33.

• At Bayern, nobody completely owned the position.

• At Marseille, Jonathan Clauss was 32.

None of this proves those clubs were preparing bids. That is not the point. The negotiation does not require White’s agent to say:

Manchester City are signing him tomorrow.

It allows him to say something more credible:

There are rich Champions League-level clubs whose playing style suits him and whose squad structure gives the move some logic.

That is very different from the generic agent line: ‘we have other options.’

What this gave White: evidence that Arsenal were not the only plausible high-level buyer for his particular skill set.

Tool #4: Put a price on reliability

Does football actually pay more for players who stay fit?

This is where the analysis becomes especially interesting.

So far we have shown that White was unusually available. But reliability only matters in a wage negotiation if clubs actually pay for it.

The real report benchmarked White’s salary against elite defenders across Europe. On public estimates, his new £150,000-a-week salary put him among the highest-paid full-backs in the Premier League.

The problem is that his per-match performance does not obviously explain that wage. My pooled performance index places White 25th of 63 Premier League full-backs. At approximately his performance level, the typical wage is closer to £80,000 per week.

So why £150,000?

One possibility is exactly what his advisers were selling:

Availability.

To test that, I used 5,815 player-seasons from Europe’s five major leagues.

First I estimated wages using the things we already know matter: performance, age, position, league, club spending power and season.

But I deliberately left availability out.

The model therefore tells us approximately what each player should earn before knowing how consistently they play.

Then I ask a simple question:

Do players with better availability systematically earn more than that benchmark?

They do.

Players available for less than 20% of league minutes over the preceding three seasons earned roughly 25% less than the model otherwise predicted.

Players in the 80–90% range earned around 17% more.

Across the full model, another 10 percentage points of past availability — approximately four league matches per season sustained over three years — is associated with about a 6% increase in wages, all else equal.

For comparison, another ten percentile points of performance are worth roughly the same.

That is a striking result.

In this sample, moving from 70% to 80% availability is associated with roughly as much pay as becoming ten percentile points better at football.

Now place White into the model.

Based on his age, performance, league, Arsenal’s wage bill and the other controls — but not his availability — the model predicts roughly £6M per year.

His new estimated contract pays around £7.8M.

About 30% more.

Players with White’s availability profile typically carry a premium of around 20% in the same analysis.

So approximately two-thirds of the premium the model cannot explain before availability is added is consistent with something the wider football labour market already pays for: being there.

This is not proof that availability caused Arsenal to give White his raise. Contract negotiations do not work like laboratory experiments.

But it tells us something useful.

His advisers were not trying to persuade Arsenal to pay for an imaginary asset. Across thousands of player-seasons, the market already appeared to value it.

What this gave White: a financial benchmark showing that reliability was not just useful on the pitch — football clubs already attached money to it.

The negotiation pack, assembled

Put the four tools together and White’s case becomes much simpler.

1. Availability: White was among the most consistently available defenders in the Premier League.

2. Season-long value: his per-match performance was good rather than elite, but combining quality with playing time pushed his contribution toward the top of the market.

3. Long-term risk: some of his strongest skills were relatively durable with age, while his centre-back experience created another possible pathway later in the contract.

4. Outside options and price: several wealthy European clubs were credible stylistic destinations, and the wider labour market appeared to pay a meaningful premium for exactly the reliability White offered.

That is much more persuasive than trying to prove Ben White was secretly the best right-back in Europe.

Because he wasn’t.

And that is what makes the case interesting.

The data scientists did not change the facts. They changed which facts the negotiation was organised around.

Instead of: ‘Ben White’s per-90 numbers are better than Player X,’ the case becomes:

Ben White is very good, almost always available, difficult to replace over an entire season, useful in multiple roles, compatible with other elite clubs and possesses an attribute the market demonstrably pays for.

That is a much stronger argument.

In March 2024, Arsenal agreed a new long-term deal reportedly taking his estimated salary from around £120,000 to £150,000 a week.

Roughly a 25% increase.

Then the asset Arsenal paid for disappeared

And this is where the story gets an unfortunate twist.

Six months after the extension, White developed a knee problem.

Two months later came surgery.

He played only 36% of Arsenal’s league minutes in 2024/25. The following season: 21%.

Since the contract was announced, his recorded injury absence reached 154 days — roughly five times the total from the preceding three and a half seasons.

At an estimated £150,000 a week, Arsenal were now paying far more for every Premier League minute White actually played.

• Before the extension, roughly £2,100 per league minute.

• 2024/25: approximately £6,300.

• 2025/26: approximately £10,600.

The economics flipped.

But that does not automatically mean Arsenal made a bad decision. That would be hindsight.

So I looked at players aged 26–28 who had played at least 85% of their club’s minutes in two consecutive seasons and asked what happened next.

Two seasons later:

• 35% remained above 85%;

• 49% landed between 50% and 85%;

• 16% fell below 50%.

Only about 4% followed White’s particular path of falling below half the available minutes in both subsequent seasons.

That is a better way to frame the outcome.

Arsenal bought an unusually strong availability record. They then received an unusually poor availability outcome.

Analytics could quantify that risk. It could not make it disappear.

The part I would steal from this case

De Bruyne’s negotiation was about proving just how expensive an elite player was to replace.

Ben White’s was subtler.

His advisers took something that usually sits in the background of football analysis — availability — and made it the centre of the commercial argument.

• First they showed it was unusual.

• Then they showed how it changed season-long contribution.

• Then they connected it to longevity, alternatives and ultimately wages.

That is the useful lesson. You do not necessarily win a negotiation by finding a metric on which you rank first. Sometimes you win by identifying the thing the standard comparison leaves out.

And there is one final implication I find interesting. If availability is part of what both sides are pricing, the obvious contract-design response is to make some compensation contingent on it.

A strong guaranteed wage plus minutes-linked bonuses would allow the player to monetise his unusually good historical record while protecting the club against precisely the tail outcome Arsenal eventually experienced.

The data does not eliminate risk.

The contract decides who carries it.

Boom — that was my reconstruction of how Ben White used data scientists during his Arsenal contract negotiation.

Unlike Kevin De Bruyne, White did not need an argument that he was one of the best players in world football.

He needed an argument for why good almost every week could be worth more than great some of the time.

And that is what makes this case so useful.

Thank you for reading until the end ❤️

I hope you enjoyed this one as much as I enjoyed putting it together.

See you next week,

Martin

P.S. This post was AI-assisted for editing and code debugging. Without it, this article would probably have missed more deadlines than Ben White missed games — once he signed the contract, of course.

P.P.S. If you liked this piece chances are you’ll also enjoy how Kevin De Bruyne himself used data scientists to secure a 30% increase at 29, or how Man City used a magical data science technique to find his replacement or what Bruno Fernandes’ data science team likely argued in his ongoing negotiations with United.

The result was a roughly 30% pay rise and a deal worth around €80M.

Bruno Fernandes now finds himself in a surprisingly similar position.

He is 32. He has around 10 months left on his Manchester United contract. He reportedly wants to stay.

And United face an uncomfortable question:

How much should they pay to retain a player approaching his mid-thirties — even if that player remains the Premier League’s most productive creator?

So I tried to recreate the exercise.

If Bruno Fernandes hired his own data team tomorrow, what would they put in front of Manchester United?

Here are the five exhibits I think they would lead with.

1. Man United depend on Bruno more than any Premier League club depends on one creator

Everyone knows Bruno Fernandes creates chances. The more interesting question is:

How much of Manchester United’s creativity actually runs through him?

Last season, Bruno broke the Premier League assists in a single season record with 21. But the more revealing number is his share of the team total.

He accounted for 46% of Manchester United’s league assists. The season before: 36%.

For context, the typical leading creator at a Premier League club accounts for roughly 16%.

Bruno is therefore operating at almost three times the normal level of creative dependence.

And this is not just a denominator trick. You could argue that his share is high simply because Manchester United create so little without him.

In 2024/25, that argument had something to it. United managed just 28 league assists, one of the five lowest totals in the division; the typical club managed 43. A big share of a small number is not hard to reach.

In 2025/26, it falls apart. United’s total rose to 46, the third-highest in the league. Bruno’s own assists more than doubled, from 10 to 21.

So his share went up while the team itself created far more. That gives his representatives a very simple negotiating line:

You are not paying for another creative midfielder. You are paying for the player through whom almost half of the team’s assists currently run.

And that immediately creates the next question.

What happens if he leaves? We’ll come back to that.

Subscribe now

2. At 32, Bruno’s availability looks more like an asset than a risk

Age will almost certainly be Manchester United’s opening counterargument.

Bruno is 32.

Any new contract would pay him through ages 33 and 34. Normally, that would immediately raise questions about availability.

Except Bruno’s record is absurd.

Since joining Manchester United, he has played between 97% and 100% of available Premier League minutes in every season.

Not once has he fallen below 97%.

And the injury record is even more striking.

Take his entire Manchester United career — from his debut in February 2020 to today, six and a half seasons.

He has lost just 27 days to injury. Over that same window, the median Premier League regular lost 184 days.

He has appeared in 233 of United’s 245 league matches. Ninety-five percent.

And he has played the third most Premier League minutes of any player in the division over the period. The two ahead of him, Jordan Pickford and James Tarkowski, lost 53 and 97 days to injury.

Among the fifty most-used players in the league since he signed, only three missed fewer days than he did: Declan Rice, Tomáš Souček and James Ward-Prowse. All three played fewer minutes. Nobody who has played more than him has missed less.

This matters because clubs normally use age to price future availability risk.

Bruno’s representatives can respond with the strongest observable evidence available:

Fine. If you think age creates an availability risk, then price the availability, because nobody in this league has more of it.

But there is another reason his minute share matters.

Manchester United are not merely paying him a large salary. They are using him like a player worth a large salary.

Across the last three seasons, he has played more Premier League minutes than any Manchester United outfield player.

Different managers. Different systems. Same result. He plays.

3. The performance data says Bruno’s decline hasn’t started

This is the most important distinction.

A 32-year-old playing every week isn’t necessarily valuable if his performance has already started falling.

Bruno’s hasn’t.

On the contrary, 2025/26 was the best creative season of his Manchester United career by several underlying measures. Across his entire United spell among Premier League attacking midfielders, he ranks around the:

• 98th percentile for expected assists

• 98th percentile for chances created

• 98th percentile for passes into the box

• 98th percentile for through balls

• 100th percentile for passes into the final third.

Any good creator will clear one or two of those. Almost nobody clears all five.

Take the 81 attacking midfielders who have played 2,700 or more Premier League minutes since 2019/20, and ask how many sit in the top 5% of the league on every one of those measures at once.

Sixty-eight of them manage none. Two manage all five.

Bruno Fernandes, and Kevin De Bruyne.

And this isn’t based on one good year. It covers more than 21,000 Premier League minutes.

Which brings us to the decline question.

Because a 32-year-old’s career numbers can hide a recent fall. So here are the same five measures scored season by season, each against that season’s attacking midfielders — the yardstick resets every year, so nothing is inherited from 2021.

The lines go up. In 2025/26 he finished in the top 3% of the league on all five, and first on three of them: chances created, through balls and passes into the final third. Whatever is going to decline hasn’t started yet.

Bruno’s side does not need to argue that he is perfect.

He isn’t.

His progression by carrying is weak. His box presence is modest. His pass completion could improve.

The interesting point is that none of those things has stopped him being one of the league’s most productive creators.

Which is what United are actually buying.

And now to the really interesting one.

4. The wage model says Bruno is not overpaid

Bruno earns an estimated €18M a year. That sounds enormous. And it is.

He is one of Manchester United’s highest-paid players and among the best-paid midfielders in Europe.

So United’s most intuitive argument writes itself:

You’re already one of the highest-paid players in the league.

True. But salary in isolation tells us almost nothing. The useful question is whether he is paid like the player he actually is.

The obvious place to start is the market. Take every attacking player in Europe’s top five leagues who played at least 900 league minutes last season, 514 of them, and compare two things: what they produce, and what they are paid.

For output, I use expected goals plus expected assisted goals per 90 minutes (xG + xAG).

Bruno produced 0.68 per 90.

• That puts him in the 95th percentile.

• His wage puts him in the 98th percentile.

He is paid like one of the best attacking players in Europe because he produces like one.

Now look at the shape of that market. For most players, wages barely move with output. The median attacking player earns about €2.4M, and nearly two-thirds earn under €3M. Then, right at the top, wages go almost vertical: the ten best-paid players take close to 12% of all the wages paid to the 514.

Bruno sits right at that edge. He is the ninth-best-paid attacking player in the top five leagues.

That is exactly where ordinary salary benchmarking starts to struggle.

To see what a conventional benchmark says anyway, I built a wage model using 1,817 players across Europe’s top five leagues. It learns what the market typically pays given attacking output (xG + xAG), share of league minutes, age, club wage bill, league and position, and it explains about 73% of the variation in wages.

Then I put Bruno’s actual numbers into it.

• Model-predicted wage: €17.2M

• Actual estimated wage: €18.0M

Bruno is paid about 5% more than the model predicts. But the typical player in the sample is also paid a little above their prediction, about 3%.

Put simply: line up all 1,817 players by how generously they are paid compared with what the model predicts for them, and Bruno lands almost exactly in the middle. Just under half of them are paid more generously than he is.

And this is where it gets interesting.

A model like this has a known blind spot. It is fitted across the whole labour market, so it smooths away the premium that superstar markets pay for the last few percentiles of quality. Look at which highly paid players it thinks are furthest above their prediction: Mbappé, Salah, Vinicius. Erling Haaland earns about 80% more than it predicts.

When I ran the same kind of model for Kevin De Bruyne’s 2021 negotiation, it priced him at around €11M against a salary of roughly €22M, and put him in the company of Messi, Neymar and Ronaldo as “overpaid”.

Bruno does not get that verdict. Even a model that tends to call elite players overpaid says he is paid about what a typical player with his profile earns.

So United’s opening line has a second half:

You’re already one of the highest-paid players in the league ... because you’re also one of its most productive and most-used players.

Among the Premier League’s highly paid creative players, Bruno ranks around: 3rd for creation and 2nd for wage.

That is not evidence of exploitation. It is something more useful in a negotiation: pay and production are broadly aligned.

And even age does less damage than you might expect. Holding his current performance and usage constant, the model’s benchmark barely moves:

• Age 32 → €17.3M

• Age 33 → €17.1M

• Age 34 → €16.8M

So United’s strongest argument is not:

Bruno is already overpaid.

The numbers don’t really support that. It is:

We don’t know whether today’s Bruno will still exist at 34.

That is a legitimate concern.

It is also a concern about the future rather than a claim about the present, which makes it an argument about how a contract is shaped rather than about what he is currently worth.

But if the wage model says he is paid fairly, where does the case for a raise come from?

From what happens to United without him, which brings us to the last point.

5. Replacing Bruno costs more than paying him

United’s cheapest move is to do nothing. His deal runs to June 2027, and United hold an option to extend it for another season on the existing terms.

So the club can reasonably ask:

Why should we pay you more, when we can keep you another year on what you earn now?

The answer is not in the wage model. It is in the league table.

So I simulated it. Twenty thousand Premier League seasons with Bruno, and twenty thousand with his creative output replaced by a lesser player.

The mechanics are simple. Take the gap between his expected goals and assists per 90 (penalties excluded, because his penalties stay with the club when he leaves) and the replacement’s. Spread it over a full season of minutes. Convert it into league points at the historical rate of about 0.63 points per goal contribution. Then let every simulated season carry the random noise real seasons have, around four points either way.

Losing him costs United roughly:

• 1.5 points against their next-best creator

• 4.4 points against a typical Premier League No.10

• 6.7 points against a league-average midfielder

Six points does not sound like much. For United, it is. Their nine-season average is 66 points; the average fifth-place finish over the same nine seasons is also 66.

• Starting from that normal level, the simulation gives United a 48% chance of a top-five finish with Bruno.

• With a typical Premier League No.10 instead: 20%.

• With a league-average midfielder: 10%.

The exact percentages are illustrative. The distance between them is not.

What those percentages are worth

For United, top five means the Champions League instead of, at best, the Europa League. That difference has a price, and most of it is public.

UEFA’s own budget for 2026/27 pays the average Champions League club about €67M, and the average Europa League club about €16M.

On top of that, United’s Adidas kit deal deducts £10M, roughly €11.5M, for every season without Champions League football.

So a Champions League season is worth at least €63M more to United than a Europa League one. At least, because the extra home matches and their ticket income are not even counted.

Multiply that by how much each replacement lowers the chance of getting there:

• United’s next-best creator: 10 points less likely to qualify, about €6M a season in expected Champions League money

• A typical Premier League No.10: 27 points less likely, about €17M

• A league-average midfielder: 37 points less likely, about €23M

That last number is more than Bruno’s entire wage.

And a real replacement has to be bought

Those three options are the cheap ones. They cost nothing to acquire.

Replacing him properly means buying a creator. So I took every attacking midfielder a Premier League club has bought for €35M or more since summer 2023, and costed each one as if United had signed them to fill his role: their wage, the fee spread over a five-year contract, and the Champions League money United would expect to lose because they create less than Bruno does.

Florian Wirtz comes to about €53M a season. Dominik Szoboszlai, €45M. Xavi Simons, €43M. Eberechi Eze, €32M. James Maddison, €26M.

Mason Mount, United’s own purchase, comes to €45M, because he creates a little over half as much per 90 as the player he would be replacing.

Only two come in at Bruno’s €18M or below: Cole Palmer and Rayan Cherki. Both were bought at 21, before the market had fully priced them, and both have out-created him per 90 (Cherki on only around 2,000 Premier League minutes so far).

That is the replacement route that works. It is also a scouting bet, not a purchase: you have to find the next one before everyone else does.

To be fair to the club, a 21-year-old keeps resale value and a 32-year-old does not, and spreading a fee over five years ignores that.

But the business point is much simpler: United are not deciding between paying Bruno €18M and saving €18M.

They are deciding between paying Bruno €18M and spending somewhere between €26M and €53M a season on somebody who, on the evidence, creates less.

Surely United can price the risk of him getting older. What they may struggle to price is the cost of losing him.

Boom, that was what Bruno’s future data science team would say if they want him to get a salary bump before his time in United is over.

I hope you enjoyed this one as much as I enjoyed creating it.

Thank you for reading until the end ❤️

See you next week,

Martin

P.S. This post was AI-assisted for editing and code debugging. Unlike Bruno, AI still hasn’t asked me for a new contract.

P.P.S. If you enjoyed this one, you might also like my much more detailed 70-page report on Kai Havertz and his upcoming contract negotiation. You can read it here. Cheers!

He earns an estimated €17M a year, the second-highest salary at Arsenal.

In 2025/26 he played 647 Premier League minutes - just 19% of the schedule.

Put those two facts next to each other and the club side of the negotiation almost writes itself: top-of-market wage, bottom-of-the-squad usage, two long absences in thirteen months. Reprice the contract.

Except the performance data says something very different.

Across his three Arsenal seasons, Havertz sits around the 89th percentile of Premier League attackers for attacking output. He is at the 95th percentile for aerial duels won. Among centre-forwards, his creation sits around the 94th percentile. And in the injury-shortened 2025/26 season, his per-90 attacking rate was the highest of his career.

Then came another twist. By the start of 2026/27, Havertz was fit again and Arsenal were starting him at centre-forward in every competitive match.

So what exactly are Arsenal supposed to price: the player, or the risk that the player is not there?

That question bothered me enough that I built a full contract-negotiation assessment around it: 69 pages of playing time, performance, wages, market value, replacement cost, simulations and contract structures.

Grab the report

Below I’ve summarized the most important points.

One important caveat before we start: this is a public demonstration case study. I do not know whether Arsenal and Havertz are negotiating, what either side wants, or what either side would actually argue. The exercise is simply: if both sides sat down tomorrow, what could each side defend with the data?

Once I stripped the report down to the few things that actually move the negotiation, the case became much simpler.

Four tools.

Let’s get into it.

Why this negotiation is harder than it looks

There are really two Kai Havertz questions, and they receive opposite answers.

How good is he when he plays? The evidence is strong.

How much football did Arsenal actually get for the money last season? Very little.

That distinction matters because football wages usually bundle those two things together. A club pays for expected performance, but it also pays for the probability that the player is available to deliver it.

When both are stable, nobody notices the distinction. Havertz forces it into the open.

1. Separate playing level from availability

Was Havertz getting worse - or simply playing less?

Start with the most uncomfortable fact for the player: 2025/26 was not merely a below-average workload season. It was an outlier.

Havertz had spent most of the previous decade as an established first-team player. Across seven seasons before 2024/25, he never missed an in-season period of five weeks or more. Then the calendar suddenly breaks twice: roughly 102 days without a competitive appearance in early 2025, then another 147-day gap later that year.

Those absences crushed his volume. His league-minute share fell from 86% in his first Arsenal season to 61%, then to 19%.

But the performance rate underneath the missing minutes did not collapse with it.

On the pooled Arsenal record, Havertz produced roughly 0.53 non-penalty expected goals plus expected assists per 90. Against the 173 Premier League attackers in the same three-season window, that places him around the 89th percentile for attacking output.

The profile is also unusual. Most high-output attackers do not win many aerials; most high-aerial attackers do not produce elite attacking output. Havertz sits in a sparsely populated corner of the market.

His profile is not perfect. Dribbling remains a genuine weakness even after I change the peer group. Some of his other apparently weak numbers disappear when he is compared with players doing the same job: progressive runs and passes into the box look ordinary against all attackers, but comfortably above average against centre-forwards.

That distinction is useful in a negotiation because it prevents both sides from arguing from the wrong denominator. The player cannot pretend the take-on weakness is not real. The club cannot price role-driven metrics as if they were evidence of decline.

What this gave Havertz: a clean way to argue that the recent problem was availability, not a deterioration in playing level.

What this gave Arsenal: the equally clean counter that rates do not pay for themselves. A top per-90 player who only plays 647 minutes still delivered only 647 minutes.

2. Price the same player twice

What should Arsenal pay for 19% availability? What about 85%?

This is where the case became much more interesting to me.

Havertz earns about €17M a year. On raw hierarchy alone, that is difficult for the player to defend after the season he just had: second-highest wage at Arsenal attached to the 20th-most league minutes in the squad.

So I built a wage model across 1,436 players in Europe’s top five leagues. It estimates what players with similar observable characteristics are typically paid, using attacking output, minutes, age, club wage bill, league and position.

The important thing is not the exact regression. It is what happens when the model prices the same footballer under two different availability assumptions.

Observed 2025/26 usage: the model values his wage at about €13.5M.

Restored starter usage: hold his performance rate constant but move him back toward an 85% share of league minutes, and the benchmark rises to about €16.7M.

His actual estimate is roughly €17M.

The model does not really disagree with Havertz’s salary. It disagrees with paying that salary for a season in which he is barely available.

That is a very different negotiation.

If the club believes 2025/26 is the new normal, €17M looks rich. If the club believes his starter usage has genuinely returned, the same model almost reproduces his existing contract.

The wage question therefore collapses into the same uncertainty as the sporting question: how much of the next contract should be priced as if Havertz is a starter?

There are other anchors, and most of them help Arsenal. Statistically similar profiles generally earn less. His realistic outside wage market is thin because only a handful of European clubs pay attackers at this level. Those are real club-side constraints.

But the range is now defensible rather than arbitrary: roughly €13M to €17M in guaranteed compensation, with the position inside that range depending mostly on the availability assumption.

What this gave Havertz: evidence that his current wage is not absurd if Arsenal expects him to be a starter again.

What this gave Arsenal: a quantitative reason not to guarantee the full amount before durability is established.

3. Price Arsenal’s walk-away option

What does waiting actually cost?

A contract negotiation is never only about the salary. The alternative matters.

For Arsenal, one alternative is to sell. But a player contract is a wasting asset: the same footballer is worth less to the selling club as expiry gets closer because the buyer is purchasing fewer years of contractual control.

Havertz is contracted to June 2028. The near-full-value sale window was summer 2026. By the time of this assessment, it had already closed.

Under the scenarios in the report, the remaining fee moves roughly like this: about €39-54M in January 2027, €25-49M in summer 2027, then zero at expiry.

That creates a subtle transfer of bargaining power.

The player does not necessarily become a worse asset as the contract shortens. But the club owns less and less of that asset. Value the club can no longer capture as a transfer fee eventually reappears on the player side through wages, signing bonuses and freedom of destination.

Waiting is therefore not neutral.

The other walk-away option is replacement. And here Arsenal run into the opposite problem: genuinely replacing the profile is expensive.

A mid-market replacement - roughly a €45M fee plus an €8M wage in the model - costs around €17M a year once the transfer fee is amortised. That is almost exactly Havertz’s current salary, for only part of the profile.

A close elite replacement pushes the annual cost toward €39M.

The internal option is obviously cheaper. Arsenal can hand the role to existing players. But after the summer departures, that also means concentrating the centre-forward depth into fewer bodies - precisely the type of squad risk a versatile player is supposed to reduce.

This does not make Havertz irreplaceable. In fact, one of the hardest points for his side to defend is that Arsenal can probably reach sporting adequacy internally.

It does make the club’s threat more complicated than “take less or we replace you.” The cheapest replacement is not like-for-like. The like-for-like replacements are not cheap.

What this gave Havertz: the argument that Arsenal cannot necessarily reduce salary cost without accepting either profile loss, squad-thinning risk or a large transfer bill.

What this gave Arsenal: the reminder that Havertz’s outside market is also narrow. Replacement is expensive for Arsenal, but moving at a €17M salary is not frictionless for the player either.

4. Stop arguing about the risk and put it in the contract

What if neither side has to be right today?

At this point the negotiation has a problem that more analysis cannot completely solve.

Arsenal can say: two serious absences in thirteen months are a new risk. Havertz can say: seven healthy seasons before them, eight absence-free months after them, and the club is already starting me again.

Both statements can be true.

A model can estimate the probability of another disrupted season. It cannot make either side certain.

So instead of forcing the disagreement into one fixed salary, I tried something more practical: make the contract move with the outcome.

The example in the report compares a fixed €17M wage with a structure around €11M guaranteed plus up to €7.5M linked to minutes and performance.

If Havertz plays only 25% of league minutes, annual pay falls to roughly €13.2M.

At 50%, around €15.4M.

At 75%, around €17.6M.

At 90%, roughly €18.5M - more than he earns today.

The point is not that those exact numbers should appear in a real Arsenal contract. I do not know the true bonus structure of his current deal, and public wage estimates capture base salary imperfectly.

The point is the mechanism.

If the disagreement is mostly about availability, the cleanest contract is one that pays more when availability materialises and less when it does not.

That allows the player to bet on his own fitness without giving up the upside. It allows the club to protect itself against the exact scenario it fears. Neither side has to concede its forecast before signing.

What this gave Havertz: a route to preserve - and potentially exceed - his current compensation if he is right about his return.

What this gave Arsenal: a way to stop paying top-of-market compensation through another lost season if the club is right about the risk.

The negotiation pack, assembled

Put the four tools together and Havertz’s contract case is much cleaner than the 69-page report makes it look.

• Playing level: his underlying performance record is intact. The best player-side evidence is not the 647-minute 2025/26 rate alone, but the three-season Arsenal record and the unusual combination of output, aerial presence and creation for a No. 9.

• Compensation: the wage model prices him at about €13.5M on last season’s actual usage and about €16.7M if starter usage is restored. The salary argument is therefore mostly an availability argument.

• Alternatives: Arsenal can replace minutes cheaply, but replacing the full profile is expensive. At the same time, every transfer window that passes reduces the fee the club can capture.

• Structure: because the key uncertainty is future availability rather than current level, the contract can price the uncertainty directly instead of forcing either side to win the forecast in advance.

The club still has a serious case.

Havertz was paid like one of Europe’s elite attackers while barely playing. Similar statistical profiles mostly earn less. His outside market at the same wage is narrow. And another long absence would weaken both his wage case and his transfer value quickly.

The player also has a serious case.

The data does not show performance decline. Arsenal have already allowed the best sale window to pass. The current team sheets say starter, not surplus. And buying a comparable external profile costs at least as much as the salary the club might be trying to reduce.

That is why I think the interesting negotiation is not really “€13M or €17M?”

It is: how much of the €17M should be guaranteed before the availability question is fully answered?

The part I would steal from this case

The Kevin De Bruyne case was about replacement cost. Ben White was about finding a dimension that ordinary per-90 analysis missed. Alex Greenwood was about fixing the frame when the market itself did not provide a useful price list.

Havertz gives us a different lesson.

Sometimes both sides are looking at the same evidence and disagree not because one of them is wrong, but because they are pricing different future states.

Arsenal can reasonably price the risk that the injuries repeat. Havertz can reasonably price the evidence that his level has not declined and his availability has returned.

The analytical mistake would be to force those two futures into one supposedly correct number.

A better answer is to identify the variable that drives the disagreement, estimate what each state is worth, and write a contract whose economics move with that variable.

That is the broader lesson I would steal: sometimes the best model does not tell you who is right. It tells you how to write the deal so you do not need to know yet.

Boom — that was the Kai Havertz contract negotiation in a nutshell.

Thank you for reading until the end ❤️

You can grab the complete version of the report below.

Grab the report

The full assessment runs to 69 pages and contains the parts I have deliberately skipped here: detailed peer pools, role adjustments, finishing, similar-player wages, market-value scenarios, with/without analysis, title simulations and the complete bilateral negotiation framework.

Talk soon,

Martin

For enquiries about tailor-made reports, drop me a line @ martin.angelov@datadealsfc.com

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.