It sorts signings well. It gets the expensive ones wrong, and we think we know why.
- 1,553 summer-2024 signings in our 14 divisions. The fifth we rated most likely to become regulars did it 48% of the time. The fifth we rated least likely: 6%.
- League moves: our shots range held for 311 of 405 players who changed league (77%). It held for moves across and down. It failed again for players stepping up from a second division.
- The 20 biggest fees: 10 right. No better than a coin flip. 7 of the 10 misses went the same way: we said he wouldn't play, he did.
The rules
Only what was known on 1 June 2024. The signing model learned from the summer 2023 window alone. Club strength is our own rating, as it stood that day. Loans made permanent don't count as signings: the player was already there.
A regular means he played at least half of his new club's league minutes after he arrived. Loaned out again in January means he wasn't a regular.
The 20 cases are the 20 biggest reported fees among those signings. Chosen by that rule, every one shown, misses included.
Sorting signings
| Our rating | We said | Became regulars |
|---|---|---|
| Least likely fifth | 12% | 6% |
| Second | 22% | 21% |
| Middle | 30% | 32% |
| Fourth | 39% | 40% |
| Most likely fifth | 54% | 48% |
The middle of the table lands within a few points. The top fifth is too confident: we said 54%, it was 48%. Across all 1,553 the ranking was right about 7 times in ten when you pair a regular with a non-regular (AUC 0.71).
League moves
For players who changed league we give a range for shots per 90 at the new club. Eight in ten should land inside it.
| Move | Inside the range | Rate | Verdict |
|---|---|---|---|
| Between top leagues | 77 of 97 | 79% | Holds |
| Down to a second division | 99 of 122 | 81% | Holds |
| Smaller top league to the big five | 25 of 30 | 83% | Holds |
| Second division to a top league | 110 of 156 | 71% | Fails |
Stepping up from a second division is still the weak spot. The average holds, the spread doesn't: some players shoot more than ever, some vanish.
The 20 biggest fees
| Player | Move | Fee | Chance he'd be a regular, we said | Share of minutes he played | Call |
|---|---|---|---|---|---|
| J. Álvarez | Manchester City to Atletico Madrid | €75m | 62% | 74% | Right |
| D. Solanke | Bournemouth to Tottenham | €64.3m | 45% | 64% | Wrong |
| L. Yoro | Lille to Manchester United | €62m | 56% | 34% | Wrong |
| Pedro Neto | Wolves to Chelsea | €60m | 16% | 67% | Wrong |
| João Neves | Benfica to Paris Saint-Germain | €60m | 38% | 61% | Wrong |
| A. Onana | Everton to Aston Villa | €59.4m | 32% | 48% | Right |
| Dani Olmo | RB Leipzig to Barcelona | €55m | 31% | 35% | Right |
| M. Olise | Crystal Palace to Bayern München | €53m | 21% | 77% | Wrong |
| T. Koopmeiners | Atalanta to Juventus | €52m | 61% | 62% | Right |
| Douglas Luiz | Aston Villa to Juventus | €51.5m | 48% | 15% | Right |
| João Palhinha | Fulham to Bayern München | €51m | 36% | 22% | Right |
| M. Ugarte | Paris Saint-Germain to Manchester United | €50m | 63% | 55% | Right |
| D. Doué | Rennes to Paris Saint-Germain | €50m | 16% | 58% | Wrong |
| M. Kilman | Wolves to West Ham | €47.5m | 55% | 98% | Right |
| A. Gray | Leeds to Tottenham | €47.2m | 34% | 51% | Wrong |
| G. Rutter | Leeds to Brighton | €46.7m | 51% | 50% | Right |
| M. de Ligt | Bayern München to Manchester United | €45m | 41% | 62% | Wrong |
| R. Calafiori | Bologna to Arsenal | €45m | 38% | 29% | Right |
| J. Zirkzee | Bologna to Manchester United | €42.5m | 53% | 41% | Wrong |
| C. Gallagher | Chelsea to Atletico Madrid | €42m | 53% | 49% | Wrong |
Two calls sit right on the line: Rutter played 50% of Brighton's league minutes, Gallagher 49% of Atletico's.
Why we got the big ones wrong
The model doesn't know a fee was paid. It learned from 1,541 signings in the summer 2023 window, and only 71 of them cost €20m or more. The rest were loans, free transfers and cheaper moves, where a step up to a bigger club usually means fewer minutes. Clubs that pay big fees usually mean to play the player: 12 of these 20 became regulars.
- Olise to Bayern: we said 21%, he played 77%.
- Pedro Neto to Chelsea: we said 16%, he played 67%.
- Doué to PSG: we said 16%, he played 58%.
The fix isn't simple. Our data feed stopped publishing fee amounts in the 2025 window, so a fee can't go into the model. And on 2024, where we do have fees, adding one helped a little, but not by enough to be sure. We tried the next best thing, the player's form and the strength of the club he left. It didn't pass on a season the model hadn't seen, so it isn't in.
Until something does, we say it out loud: for paid moves into the clubs at the top of our ratings, our number runs low. By 4 points in 2024. By 14 in 2025.
What's next
The next version adds competition for the shirt: had the club just lost a starter in his position? Same rules as everything else here. Written down before the test, scored on a season it hasn't seen, published either way.
Data sources
- API-Football: transfers and reported fees, player season statistics, league fixtures and results.
- Club strength: our own rating, built from those results.
Want this run on a player you're working on? We'll do the first one free. How it works →