Hand Manchester City's next season to Scott Parker and the model still gives him 71 points.
Parker finished bottom of a 23-manager field we ran through the same City squad, the same 38 fixtures, the same everything. Bottom of that pile is still a top-four season. Pep Guardiola came first on 75.62 expected points, which is the result you would guess and the least interesting thing in the data. A full season of the best manager in the field is worth 4.48 points more than a full season of the worst one.
Enzo Maresca is the name that matters here, because he is the one actually taking the job. He ran City's Elite Development Squad, he sat on Guardiola's first-team staff, and he has spent his time at Chelsea building a version of the same patient, possession-first shape he learned at this club. The model puts him fifth of 23. It also says fifth and first are close to the same place.
What we actually changed
The setup is deliberately boring. City's team rating stays where it is, and so does the fixture list, home and away against the other 19 clubs in next season's Premier League field. We change the coach the model thinks is standing on the touchline, then rerun the season from scratch, once per manager.
Holding the squad still is what a first season actually looks like. A new manager walks into a dressing room that is already settled, with the spine picked, the contracts signed and the habits set, and he gets one window at most to change any of it. Everything this test measures is the year before a manager has had the chance to build anything.
For every fixture the model returns a win, draw and loss probability. Expected points are three times the win probability plus the draw probability, which is the average number of points that fixture is worth before anyone kicks off.
How far does the table actually move?
Not far. Guardiola tops the list on 75.62, Maresca is fifth on 74.46, Arne Slot sits in the middle of the field and Parker is last. The four names ahead of Maresca are covered by 0.58 points between them.
4.48 points separate top from bottom
These ranks come from one frozen model run and carry no confidence intervals. Open the table for all 23 coaches.
Guardiola and Unai Emery are separated by 0.01. Maresca sits No. 5, 1.16 points behind Guardiola.
All coach projections
| Rank | Coach | PL matches | Expected points | Percentile |
|---|---|---|---|---|
| 1 | Pep Guardiola | 48 | 75.62 | 100.0% |
| 2 | Unai Emery | 50 | 75.60 | 95.5% |
| 3 | Mikel Arteta | 51 | 75.34 | 90.9% |
| 4 | Eddie Howe | 50 | 75.04 | 86.4% |
| 5 | Enzo Maresca | 42 | 74.46 | 81.8% |
| 6 | Nuno Espírito Santo | 50 | 73.89 | 77.3% |
| 7 | Marco Silva | 49 | 73.87 | 72.7% |
| 8 | Vítor Pereira | 38 | 73.84 | 68.2% |
| 9 | Fabian Hürzeler | 50 | 73.76 | 63.6% |
| 10 | David Moyes | 20 | 73.54 | 59.1% |
| 11 | Régis Le Bris | 30 | 73.50 | 54.5% |
| 12 | Arne Slot | 50 | 73.49 | 50.0% |
| 13 | Andoni Iraola | 50 | 73.43 | 45.5% |
| 14 | Oliver Glasner | 48 | 73.32 | 40.9% |
| 15 | Thomas Frank | 23 | 73.12 | 36.4% |
| 16 | Daniel Farke | 30 | 72.89 | 31.8% |
| 17 | Kieran McKenna | 19 | 72.86 | 27.3% |
| 18 | Graham Potter | 19 | 72.86 | 22.7% |
| 19 | Keith Andrews | 30 | 72.61 | 18.2% |
| 20 | Ruben Amorim | 38 | 72.58 | 13.6% |
| 21 | Ange Postecoglou | 19 | 72.55 | 9.1% |
| 22 | Ruud van Nistelrooy | 19 | 71.48 | 4.5% |
| 23 | Scott Parker | 31 | 71.14 | 0.0% |
A table has to put somebody first. Read the scale underneath the names and the ordering stops meaning very much. None of these ratings carry uncertainty intervals either, so a few hundredths of a point is a tie, which makes the top four a tie.
The ranking hides the overlap
The ranking squeezes a manager into a single number. Run the season 100,000 times and you get a shape instead, and the shapes sit on top of each other. Maresca's middle 80% covers 64 to 84 points.
Guardiola's middle 80% runs 65 to 86 points
Each Guardiola season is paired with a Maresca season using the same fixture-level randomness. The peer curve mixes every peer coach with equal weight.
Distribution summary
| Scenario | Mean | Median | P10 | P90 |
|---|---|---|---|---|
| Guardiola | 75.6 | 76 | 65 | 86 |
| Maresca | 74.4 | 75 | 64 | 84 |
| Peer mixture | 73.3 | 73 | 63 | 83 |
A poor year under Guardiola looks a lot like a strong year under most of the league. City's floor is doing far more work here than City's manager.
Same season, same luck, different manager
The fair objection is that two wide distributions will always overlap and that tells you nothing. So we paired them. Every Guardiola season runs against a Maresca season using the same random draw for the same fixture. If City are unlucky at Goodison in one branch, they are unlucky at Goodison in the other. Whatever survives that is the coach.
In 44% of paired seasons the two branches finish on exactly the same points total. Guardiola's branch comes out ahead in 54% of them and Maresca's in 1.6%. The median gap across a full 38 matches is 1 point.
So where does Guardiola's edge come from?
The fixture-level view is the one that settled it for us. The most coach-sensitive match of the 38 is at home to Manchester United, worth 0.12 expected points. The least sensitive is away at Arsenal, at 0.001, because that fixture is hard for anybody.
Guardiola gains points at both venues
Expected points added over Maresca across City's 19 home and 19 away matches.
All 38 fixture differences
Values below are Maresca expected points minus Guardiola expected points, so negative figures favour Guardiola.
| Opponent | City home | City away |
|---|---|---|
| Arsenal | -0.0061 | -0.0011 |
| Aston Villa | -0.0231 | -0.0028 |
| Bournemouth | -0.0292 | -0.0035 |
| Brentford | -0.0562 | -0.1057 |
| Brighton & Hove Albion | -0.0146 | -0.0157 |
| Chelsea | -0.0187 | -0.1151 |
| Coventry City | -0.0312 | -0.0091 |
| Crystal Palace | -0.0324 | -0.1026 |
| Everton | -0.0148 | +0.0087 |
| Fulham | -0.0160 | -0.0140 |
| Hull City | -0.0492 | -0.0081 |
| Ipswich Town | -0.0266 | -0.0106 |
| Leeds United | -0.0348 | -0.0864 |
| Liverpool | -0.0451 | -0.0304 |
| Manchester United | -0.1230 | +0.0177 |
| Newcastle United | -0.0333 | -0.0120 |
| Nottingham Forest | -0.0499 | -0.0042 |
| Sunderland | -0.0408 | +0.0050 |
| Tottenham Hotspur | -0.0328 | +0.0104 |
What the test cannot see
Guardiola has been at City for a decade. He signed players, built the patterns and set the standard the squad now holds itself to, and every bit of that has already been absorbed into City's team rating. When the model prices the coach term at around a point, it is pricing what is left over after everything Guardiola built has already been counted as Manchester City. So the coach term measures what remains to be lost once the building is finished, which is a very different question from how good the builder was.
That also fixes the horizon at one year. The answer here is built on the squad Maresca inherits rather than the one he would eventually assemble. A manager who lasts gets windows, and with them he stops nudging a coach term and starts rebuilding the team this test is holding still.
The job Maresca is walking into
Even though Guardiola has the highest ranking in our data, Man City depth seems to be robust enough and ready for the Maresca takeover. He is one of the best choices.
Fifth of 23 is a good rating for a manager this early in his career at the top. It looks better once you notice that the four names above him are all in work and the gap to them is about a point.
The risk sits in the part the model cannot price: whether a squad keeps its level when the person who set that level is no longer the one enforcing it. That shows up in City's team rating long before it shows up in anyone's coach rating, and the team rating is the number worth watching from August.