★ Moneyball Original · Opinion
Manchester United have a 22% chance of relegation. Is it real?
By Moneyball Editorial · published · updated
The supercomputer has spoken: Manchester United have a 22% chance of relegation. The figure from Opta's model went round the world after United lost 2-0 at Chelsea on 31 October, a third straight defeat and nine games without a win. Twenty times English champions, three times champions of Europe, and one in five chances of the Championship. It sounds absurd. Is it?
Moneyball does not take anyone's number on trust, not even a supercomputer's. So we ran our own. Ten thousand simulated seasons, every match played from the chances each club really creates and allows, read with twelve Moneyball methods and the Stats Centre. In our model Manchester United go down in 10% of them. This column is the case for and against that number.
How did it come to this?
United sit 18th on 8 points from 10 matches (1 won, 5 drawn, 4 lost), goal difference −5. Behind the table is a familiar story: a squad rebuilt again and again, a manager under pressure, and an ownership, INEOS and Sir Jim Ratcliffe, that promised to fix the football first.
Results are only the surface. A club is relegated because, over 38 matches, it creates fewer chances than it allows. The first question for Moneyball is therefore not "how many points?" but "how good are United really?".
Are United as bad as their points?
| Measure | Manchester United | Rank of 20 |
|---|---|---|
| League position (points) | 8 pts from 10 | 18th |
| Expected points (from xG) | 14.4 from 10 | 9th |
| Pi-rating (goal margin vs average) | −0.09 | 17th |
| Strength after shrinkage (xG margin per match) | +0.14 | 8th |
| Projected final points | 49 | 13th |
On expected points, the measure of what United's chances usually earn, they would have 14.4, not 8. That is 6.4 points fewer than their chances deserve: United have been unlucky, and the regression method expects some of those points to come back. Their xG margin is +0.26 a match, 8th in the league after Bayesian shrinkage.
Finishing and goalkeeping. United have scored 14 from 17.6 xG (−3.6) and conceded 19 from 15.0 xG against (−4.0 kept out). The finishing has cost them goals the chances should have brought: that usually corrects itself. The goalkeeping has let in more than the shots faced deserve: that too usually turns.
10,000 seasons: the Moneyball verdict
We played the rest of the Premier League 10,000 times. Every club plays every match it has left at its own strength (its xG margin per match, shrunk toward the league average while the sample is small), each result drawn from the Dixon-Coles score matrix. No reputation, no wage bill, no badge: only the chances.
| In 10,000 simulated seasons | Manchester United |
|---|---|
| Relegated (18th to 20th) | 10% |
| Finish in the top half | 36% |
| Finish in the top four | 3% |
| Most likely finishing place | 12th |
| Typical final points (median) | 49 |
| Points the 17th-placed club usually ends on | 41 |
The answer: on today's numbers United are in danger, though more likely to survive. Our model is kinder than the supercomputer: United's underlying chances are better than their results, and the simulation gives them credit for it.
The relegation fight, measured
| Club | Pts | xPts | Pi | Relegated in 10,000 seasons |
|---|---|---|---|---|
| Crystal Palace | 7 | 8.8 | −0.20 | 88% |
| Coventry City | 9 | 9.6 | −0.13 | 51% |
| Ipswich Town | 11 | 11.5 | −0.07 | 34% |
| Hull City | 11 | 9.7 | −0.01 | 29% |
| Newcastle United | 12 | 10.4 | −0.04 | 26% |
| Aston Villa | 13 | 10.8 | −0.02 | 21% |
| Sunderland | 8 | 15.0 | −0.14 | 15% |
| Manchester United | 8 | 14.4 | −0.09 | 10% |
The clubs most likely to go down are rarely the ones with the fewest points today. They are the ones whose chances are worst: a side low in the table but strong on xG climbs, a side high in the table but weak on xG falls.
The bottom of the table
| # | Club | Pts | xPts | Pts − xPts | Projected |
|---|---|---|---|---|---|
| 15 | Fulham | 10 | 15.1 | −5.1 | 52 |
| 16 | Nottingham Forest | 9 | 16.0 | −7.0 | 52 |
| 17 | Coventry City | 9 | 9.6 | −0.6 | 40 |
| 18 | Manchester United | 8 | 14.4 | −6.4 | 49 |
| 19 | Sunderland | 8 | 15.0 | −7.0 | 47 |
| 20 | Crystal Palace | 7 | 8.8 | −1.8 | 33 |
What staying up takes
To reach the 41 points that usually keep the 17th-placed club up, United need 33 more from 28 matches: 1.18 points a match. So far they have taken 0.80 a match; at their shrunk strength the model expects 1.47. On their underlying level they should get there comfortably, which is why the simulation keeps them up most of the time.
Home and away tell the rest: 1.00 points a match at Old Trafford, 0.60 away. Relegation fights are won at home.
How United play
| Per match, season | Manchester United |
|---|---|
| Possession | 55.0% (6th of 20) |
| Pass accuracy | 85.5% (5th of 20) |
| Pressure Index | 58.3 (5th of 20) |
| Shots | 19.3 (1st of 20) |
| Big chances | 1.9 (6th of 20) |
| xG for | 1.76 (4th of 20) |
| xG against | 1.49 (12th of 20) |
| Clean sheets | 0.0 (19th of 20) |
The ranks are against the whole Premier League. Possession says who had the ball; expected goals say what each side did with it.
The players who must save them
United's best-rated players this season across the Premier League and Europe (270 minutes or more). Each name opens his Stats Centre page.
| Player | Minutes | Rating | Goals | Assists | xG |
|---|---|---|---|---|---|
| Bruno Fernandes | 1135 | 7.56 | 6 | 3 | 5.39 |
| Matheus Cunha | 947 | 7.28 | 4 | 2 | 3.09 |
| Kobbie Mainoo | 723 | 7.13 | 0 | 1 | 0.37 |
| Youri Tielemans | 899 | 7.05 | 1 | 2 | 1.02 |
| Luke Shaw | 427 | 6.92 | 0 | 1 | 0.05 |
| Bryan Mbeumo | 1231 | 6.92 | 4 | 1 | 4.31 |
| Marcus Rashford | 420 | 6.87 | 0 | 0 | 1.22 |
| Mason Mount | 322 | 6.82 | 0 | 0 | 0.96 |
The new arrivals
How the players United signed are rated now against their last season elsewhere: on average −0.05.
| Player | Came from | Rating then | Rating now | Change |
|---|---|---|---|---|
| Carlos Baleba | Brighton & Hove Albion | 6.65 | 6.70 | +0.05 |
| Andrey Santos | Chelsea | 6.76 | 6.73 | −0.03 |
| Youri Tielemans | Aston Villa | 7.15 | 7.05 | −0.10 |
| Marcus Rashford | FC Barcelona | 6.99 | 6.87 | −0.12 |
The case against panic
- The season is long. With 28 matches left, a run of form changes everything. The simulation already counts that: it is why the chance is 10%, not 100%.
- Chances repeat, luck does not. United's points are below their chances, and that gap usually closes.
- The squad is better than the table. A Premier League squad of internationals rarely finishes in the bottom three. Rarely is not never.
The Moneyball view
A 22% headline is a warning, not a sentence. On our numbers Manchester United go down in 10% of 10,000 seasons and most often finish 12th. That is a club in trouble, not yet a club going down. This column updates itself after every round: check back to see which way the numbers move.
- Expected goals (xG): the quality of every chance, so a team is judged on what it created and allowed, not only on what went in.
- Expected points: each match's xG run through a Dixon-Coles score matrix (the Poisson model with the low-score correction the Match Board uses) to give the points a performance usually earns.
- Pi-ratings: a rating for every club at home and away, moved after each result by how far the goal margin beat or missed what the ratings expected.
- Bayesian shrinkage: a club's xG margin is pulled toward the league average as if it had 8 average matches more, so a hot month is not mistaken for a new level.
- Projection: the shrunk strength played out against an average opponent for every match left, added to the points already won.
- Monte Carlo season simulation: the rest of the league played 10,000 times, every match drawn from the Dixon-Coles matrix at each club's shrunk strength, for the chance of every finishing place.
- Finishing regression: goals scored against the xG behind them. A big gap rarely lasts a season.
- Goalkeeping regression: goals conceded against the xG faced, the other end of the same test.
- Home and away split: points per match on each side of the fixture list.
- League ranks on Sportmonks' team figures: possession, passing, pressure, chances, duels.
- Player ratings per 90, from the same Sportmonks data that runs Moneyball Scout.
- Points-to-safety line: from the 10,000 seasons, the points the 17th-placed club usually ends on, and the points a match United need to reach it.
Analysis by Moneyball Editorial. The 22.16% figure is Opta's, as reported after United's defeat at Chelsea on 31 October 2026; the club's history is the published record. Every other figure is read from the Moneyball Stats Centre (Sportmonks data) and updates itself with the data, and the sentences around the figures change with them.
Manchester United in the Stats Centre The Premier League table
Questions
- Can Manchester United be relegated this season?
- Yes, on the numbers it is possible. In Moneyball's 10,000 simulations of the rest of the Premier League, United finish in the bottom three 10% of the time. Opta's supercomputer put the figure at 22.16% after their defeat at Chelsea.
- When were Manchester United last relegated?
- In 1974. United went down from the First Division at the end of 1973/74 and came straight back up a year later. They have been in the top flight ever since.
- How many points do Manchester United need to stay up?
- In the simulations the 17th-placed club usually finishes on about 41 points. United have 8 from 10 matches, so they need about 33 more from 28: 1.18 a match.
- Who is Manchester United's manager?
- Michael Carrick, under the club's football leadership led by INEOS and Sir Jim Ratcliffe, who completed a minority stake of about 27%.
- Where do the numbers come from?
- Sportmonks match, team and player data, read by the Moneyball engine and refreshed with every update of the site. The figures in this article change with the data.