- 3 pts from the last 7 games (all competitions). Goals 6 for, 13 against.
- −0.57 xG difference per game over those 7: 1.07 created, 1.64 conceded.
- 1.9 v 3.0 big chances created against conceded per game since 13 Sep.
- 6 pts dropped from winning positions in the league (Everton, Leeds, Spurs).
United have been out-created in every one of their last seven games since 13 September and have taken 3 points from them. The early league numbers looked fine because of one 5-2 win over Ipswich. The same few problems repeat: conceding to fast starts and late equalisers, wasting leads, and not turning territory into chances. Selection has also been unsettled: 23 different starters, an average of four changes between games, and eight players listed as injured at Como. Some of the poor results are bad luck, and last season's good run was better than the chances behind it.
The 11 games
League (PL), Champions League (UCL) and the Carabao Cup. xG is the expected goals each team's chances were worth. xPts is the points a result would be worth on average given both teams' xG (a rough Poisson estimate that treats each team's goals as independent and ignores game state).
| Date | Comp | Opponent | Result | xG | xGA | xGD | Big ch. | Box touches | xPts |
|---|---|---|---|---|---|---|---|---|---|
| 22/08 | PL | Hull (A) | L 0-2 | 1.83 | 1.01 | +0.82 | 2/3 | 41/11 | 1.93 |
| 30/08 | PL | Ipswich (H) | W 5-2 | 4.77 | 2.06 | +2.71 | 9/4 | 44/13 | 2.49 |
| 06/09 | PL | Everton (A) | D 2-2 | 0.69 | 1.01 | -0.32 | 1/1 | 26/32 | 1.07 |
| 10/09 | UCL | Sabah (H) | W 4-0 | 3.54 | 0.84 | +2.70 | 5/2 | 42/12 | 2.67 |
| 13/09 | PL | Man City (H) | L 0-1 | 1.00 | 1.10 | -0.10 | 1/2 | 28/16 | 1.27 |
| 16/09 | Cup | Brighton (H) | L 2-3 | 1.20 | 2.14 | -0.94 | 2/4 | 17/40 | 0.83 |
| 20/09 | PL | Fulham (A) | D 1-1 | 1.60 | 1.62 | -0.02 | 2/3 | 43/20 | 1.37 |
| 10/10 | PL | Spurs (H) | D 1-1 | 0.80 | 1.62 | -0.82 | 1/4 | 23/32 | 0.80 |
| 13/10 | UCL | Atlético (A) | L 0-2 | 0.98 | 1.47 | -0.49 | 4/2 | 14/17 | 1.02 |
| 18/10 | PL | Leeds (A) | D 1-1 | 0.92 | 1.41 | -0.49 | 2/2 | 14/21 | 1.01 |
| 21/10 | UCL | Como (A) | L 1-4 | 0.99 | 2.11 | -1.12 | 1/4 | 18/20 | 0.70 |
Record: 2 wins, 4 draws, 5 defeats; 17 goals for, 19 against; 10 points from 11 competitive games, against about 15 implied by xG. Excluding the two heaviest wins (Ipswich, Sabah), the nine other games give 8 goals for, 17 against, and an xG difference of −0.39 per game.
1. The chance difference turned negative
United won the xG battle in only 3 of the 11 games: Hull, Ipswich and Sabah. Since the Man City game (13 Sep) they have been behind on xG in all seven, although City and Fulham were within 0.1.
Bar chart of Manchester United's xG difference in each of 11 games; only Hull, Ipswich and Sabah are positive.
The profile is a team that gets the ball forward and then does little with it: 31 box touches per league game against 21 for opponents, but only 1.5 big chances per league game outside Ipswich, against 2.5 conceded. Against Hull they had 72% possession, 21 shots and 41 touches in the box for 0.60 xGOT (the quality of shots on target). Against Leeds they had 39% possession and fewer box touches (14) than their opponent (21).
2. When the goals arrive
Fast starts have been decisive. Atlético scored in the first minute and Como in the second, and United lost both Champions League away games. In the league the damage comes late: United led at Everton, Leeds and Spurs and drew all three. Two of those draws were lost to goals in the final minutes (Everton 90+6, Spurs 90+2), and in the Cup a 2-0 lead by the tenth minute turned into a 2-3 defeat.
Timeline of the minute of every goal scored and conceded by United in 11 games, with red cards marked.
Two games against ten men
Man City played from the 23rd minute with ten after Foden's red card and won 0-1, with 1.00 xG for United against 1.10 for City over the full game. Spurs went down to ten in the 78th minute with United ahead, and Solanke equalised in the 92nd. That is 1 point from 2 games with a numerical advantage.
Set pieces: a watch item
In the league United create 0.22 xG per game from set plays and concede 0.45. Leeds got 1.19 of their 1.41 xG from set pieces, and Hull got 0.94 of their 1.01. Across all 11 games the gap is smaller (0.21 against 0.30), so this is a pattern to watch, not yet a proven weakness.
Europe has been uneven
The 4-0 over Sabah (3.54 xG to 0.84) was the only comfortable game. At Atlético United were 0-1 down and went to ten men on 74 minutes. At Como they conceded four, and Como's 2.11 xG to 0.99 was not close to a freak result. Both away trips ended without a point.
A small league sample
In the league only 7 games have been played. United are 7th on xG difference out of 20, so the table is not a straightforward guide to quality yet. The sign to watch is the trend since mid-September, not the total.
3. Pressing and territory
United's pressing this season is average, and it depends on the venue. Their PPDA (passes the opponent is allowed to make for each defensive action United take high up the pitch; lower means more pressing) is 11.3 in the league, 12th of 20 against an average of 11.4. At home it is 7.8, a level that would rank among the three most aggressive teams in the league over a full season. Away it is 16.7: in the last three away games (Everton, Fulham, Leeds) United made only 14, 14 and 17 of those defensive actions while the opponent made 222, 281 and 338 passes there.
| Games | PPDA (rank) | PPDA faced | Deep completions for | Deep completions against | |
|---|---|---|---|---|---|
| 2025/26 before Carrick | 21 | 9.6 (5th) | 11.1 | 7.4 (5th) | 6.0 (5th best) |
| 2025/26 under Carrick | 17 | 12.5 (15th) | 11.6 | 7.6 (7th) | 8.2 (16th) |
| 2026/27 league so far | 7 | 11.3 (12th) | 12.1 | 9.7 (4th) | 4.6 (3rd best) |
League games only. PPDA faced is the same measure for United's opponents (how hard they press United). Deep completions are passes completed within about 20 yards of goal, per game. Ranks are out of 20, where 1st is the most pressing, the most deep completions and the fewest conceded.
Bar chart of United's PPDA in 7 league games; the home games are low (more pressing) and the away games at Everton, Fulham and Leeds are high.
Territory is the other side of it. United complete 9.7 deep passes per game, 4th in the league against an average of 7.0, and allow only 4.6, 3rd best. That control faded in the last two league games: 8 and 4 deep completions, while conceding 9 and 5. Opponents also pressed United hard in those games (PPDA faced of 8.3 against Spurs and 6.5 at Leeds, against 12.1 for the season), and at Leeds United had 39% possession and completed 75% of their passes. The pattern is a team that controls the game when it presses and plays at home, and loses control once opponents press it and it defends deep.
4. Last season's run overstated the team
United finished third in 2025/26 with 71 points. Carrick took over for the City game on 17 January and the next 17 league games brought 39 points (2.29 per game). Over the same games the chances were worth 1.54 points per game. The chances created were also weaker than in the 21 games before him (Amorim for almost all of them). The way United defended changed as well: before Carrick they were the fifth most intense pressing team (PPDA 9.6) and allowed 6.0 deep completions per game, and under Carrick PPDA rose to 12.5 (15th) and the deep completions allowed to 8.2 (16th).
| 2025/26 league | Games | Pts/game | xPts/game | xG/game | xGA/game | xGD/game |
|---|---|---|---|---|---|---|
| Before Carrick (games 1–21) | 21 | 1.52 | 1.67 | 1.83 | 1.28 | +0.54 |
| Carrick (games 22–38) | 17 | 2.29 | 1.54 | 1.59 | 1.27 | +0.32 |
| Whole season | 38 | 1.87 | 1.61 | 1.72 | 1.28 | +0.44 |
Line chart of rolling eight-game points won against points implied by xG in 2025/26; points won pull ahead after Carrick's first game.
Three readings fit these numbers, and the data cannot separate them: finishing and goalkeeping above the level of the chances, a run of good luck, or a team that was better at winning games than at creating chances. A single season of goals against xG cannot support a verdict on finishing skill. What it does say is that the 2.29 points per game was unlikely to last. This season the results have swung the other way, 10 points against about 15 implied by xG, but the underlying chance difference has also dropped, so it is not only bad luck.
5. The players
Player data comes from lineups, minutes, ratings and individual shots for all 11 games. Ratings come from a rating model and tend to favour attackers, so treat them as context. Per-90 numbers need enough minutes, so players below 150 minutes are left out of the output table.
Selection: no settled team
United used a 4-2-3-1 in every game, so the shape has not changed. The players have: 23 different starters in 11 games and 0 to 8 changes between games (an average of 4). Only Tielemans has started every game. Mbeumo, Fernandes and Lammens started 10 of 11 and were left out of the Cup XI. The Cup XI changed 8 players from the City game according to the lineups (press reports said nine) and 8 changed again for Fulham.
The centre-back pairing has changed constantly. Seven different combinations of centre-backs started the 11 games, and Maguire and Martínez were in the same XI six times. In the four games where they were the only two centre-backs United conceded six goals from 5.8 xG.
Eight players are listed as injured at Como: Heaven, Ugarte, Martínez, Shaw, Mainoo, Šeško, Rashford and Dorgu. Martínez was substituted at half-time at Leeds three days earlier. The injury list is only shown on the Como team sheet, so how many of them missed earlier games is unknown. At Como the starting defenders de Ligt (5.1) and Mazraoui (4.6) had the two lowest ratings of the XI.
| Player | HUL (L) | IPS (W) | EVE (D) | SAB (W) | MCI (L) | BHA (L) | FUL (D) | TOT (D) | ATM (L) | LEE (D) | COM (L) | Min | Avg rating |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bryan Mbeumo | 90 | 90 | 90 | 72 | 90 | 24 | 90 | 90 | 90 | 90 | 86 | 902 | 7.23 |
| Senne Lammens | 90 | 90 | 90 | 90 | 90 | · | 90 | 90 | 90 | 90 | 90 | 900 | 6.92 |
| Bruno Fernandes | 90 | 90 | 90 | 90 | 90 | 17 | 90 | 90 | 80 | 90 | 67 | 884 | 7.41 |
| Youri Tielemans | 67 | 90 | 90 | 45 | 67 | 90 | 90 | 59 | 66 | 84 | 90 | 838 | 7.37 |
| Diogo Dalot | 10 | 74 | 84 | · | 90 | 90 | 90 | 31 | 90 | 90 | 90 | 739 | 6.95 |
| Harry Maguire | 90 | 90 | 90 | · | 90 | · | 90 | 90 | 10 | 90 | 90 | 730 | 7.03 |
| Lisandro Martínez | · | 90 | 90 | 90 | 90 | · | 90 | 90 | 90 | 45 | · | 675 | 6.75 |
| Matheus Cunha | 80 | 80 | 70 | 90 | 90 | 10 | 90 | 89 | 10 | 6 | 45 | 660 | 7.10 |
| Kobbie Mainoo | 23 | 80 | 84 | 64 | 90 | 24 | 80 | 90 | · | · | · | 535 | 7.39 |
| Marcus Rashford | 45 | 80 | 70 | · | 83 | 80 | 77 | 1 | 66 | · | · | 502 | 7.10 |
| Luke Shaw | 90 | 73 | 89 | · | · | · | 84 | 89 | · | · | · | 425 | 7.14 |
| Leny Yoro | · | 16 | 6 | 90 | · | 90 | · | 59 | 74 | 90 | · | 425 | 6.79 |
| Noussair Mazraoui | 80 | 17 | 1 | 90 | · | 90 | · | · | · | 45 | 90 | 413 | 6.34 |
| Andrey Santos | 67 | 10 | 6 | 26 | · | 66 | · | · | 90 | 90 | 45 | 400 | 6.73 |
| Ayden Heaven | 90 | · | · | · | · | 90 | · | 1 | 80 | · | · | 261 | 6.15 |
| Mason Mount | · | · | · | 45 | · | 73 | 10 | · | 24 | 16 | 90 | 258 | 6.71 |
| Patrick Dorgu | 45 | 10 | 20 | 72 | 90 | · | 6 | · | · | · | · | 243 | 6.89 |
| Amad Diallo | · | · | · | · | · | · | · | 59 | 24 | 90 | 45 | 218 | 7.16 |
| Carlos Baleba | · | · | · | · | · | · | · | 31 | 66 | 74 | 45 | 216 | 6.84 |
| Benjamin Šeško | · | · | 20 | 64 | 23 | 66 | · | · | · | · | · | 173 | 6.80 |
| Joshua Zirkzee | · | · | · | 26 | 7 | 24 | 13 | 31 | 24 | · | 23 | 148 | 6.40 |
| Shea Lacey | 10 | · | · | 18 | · | 66 | · | · | · | · | 4 | 98 | 6.71 |
| Karl Darlow | · | · | · | · | · | 90 | · | · | · | · | · | 90 | 5.25 |
| Matthijs de Ligt | · | · | · | · | · | · | · | · | · | · | 79 | 79 | 5.09 |
Minutes played by each player with 60 or more minutes, in order of minutes. Solid cells are starts, pale cells are substitute appearances, a dot means not used or unavailable. The letter is the United result. Games run left to right from Hull (22 Aug) to Como (21 Oct): HUL Hull, IPS Ipswich, EVE Everton, SAB Sabah, MCI Man City, BHA Brighton (Cup), FUL Fulham, TOT Spurs, ATM Atlético, LEE Leeds, COM Como.
Output: three players carry the attack
Mbeumo, Fernandes and Cunha produced 9.5 of the 19.3 xG credited to individual players (49%). Mbeumo has 4 goals from 4.41 xG. Fernandes has 4 goals from 3.38 xG and leads the team in chances created (25) and expected assists (2.02). Cunha has 2 goals from 1.71 xG. Rashford has no goals from 1.93 xG in 502 minutes, which is too little to judge his finishing.
The forward positions look like the unsettled part. Cunha has the minutes (660) at 0.23 xG per 90. Šeško (0.58) and Zirkzee (0.90) have higher rates in only 173 and 148 minutes, too few to rank them, and Šeško is now injured.
| Player | Apps (starts) | Min | Avg rating | G | A | xG | xA | Shots /90 | Key passes /90 | Box touches /90 | Def. actions /90 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Bryan Mbeumo | 11 (10) | 902 | 7.23 | 4 | 1 | 4.41 | 1.38 | 2.8 | 1.6 | 4.3 | 3.4 |
| Bruno Fernandes | 11 (10) | 884 | 7.41 | 4 | 1 | 3.38 | 2.02 | 3.2 | 2.5 | 2.6 | 4.9 |
| Youri Tielemans | 11 (11) | 838 | 7.37 | 0 | 3 | 0.82 | 1.52 | 1.4 | 1.8 | 1.5 | 8.5 |
| Diogo Dalot | 10 (8) | 739 | 6.95 | 0 | 0 | 0.31 | 0.41 | 1.2 | 1.9 | 2.2 | 5.7 |
| Harry Maguire | 9 (8) | 730 | 7.03 | 0 | 0 | 0.38 | 0.49 | 1.5 | 0.4 | 2.0 | 7.4 |
| Lisandro Martínez | 8 (8) | 675 | 6.75 | 1 | 0 | 1.13 | 0.56 | 1.1 | 0.3 | 1.1 | 6.5 |
| Matheus Cunha | 11 (8) | 660 | 7.10 | 2 | 2 | 1.71 | 1.07 | 2.9 | 1.5 | 5.5 | 4.5 |
| Kobbie Mainoo | 8 (6) | 535 | 7.39 | 0 | 1 | 0.30 | 0.67 | 1.0 | 1.2 | 0.7 | 12.8 |
| Marcus Rashford | 8 (6) | 502 | 7.10 | 0 | 0 | 1.93 | 0.47 | 3.0 | 2.5 | 7.7 | 3.4 |
| Luke Shaw | 5 (5) | 425 | 7.14 | 0 | 1 | 0.03 | 0.59 | 0.4 | 1.1 | 1.9 | 7.6 |
| Leny Yoro | 7 (5) | 425 | 6.79 | 0 | 0 | 0.07 | 0.08 | 0.2 | 0.2 | 0.4 | 11.0 |
| Noussair Mazraoui | 7 (4) | 413 | 6.34 | 0 | 0 | 0.39 | 0.16 | 1.1 | 0.9 | 1.7 | 7.4 |
| Andrey Santos | 8 (5) | 400 | 6.73 | 0 | 0 | 0.00 | 0.31 | 0.0 | 0.4 | 1.1 | 9.0 |
| Ayden Heaven | 4 (3) | 261 | 6.15 | 0 | 0 | 0.12 | 0.06 | 0.7 | 0.3 | 1.0 | 4.8 |
| Mason Mount | 6 (2) | 258 | 6.71 | 1 | 0 | 0.39 | 0.16 | 1.7 | 0.7 | 2.1 | 7.3 |
| Patrick Dorgu | 6 (3) | 243 | 6.89 | 0 | 2 | 0.88 | 0.72 | 3.0 | 2.2 | 8.5 | 10.4 |
| Amad Diallo | 4 (2) | 218 | 7.16 | 0 | 1 | 0.25 | 0.23 | 2.9 | 1.2 | 5.4 | 7.0 |
| Carlos Baleba | 4 (2) | 216 | 6.84 | 0 | 0 | 0.12 | 0.06 | 1.7 | 1.2 | 1.2 | 10.4 |
| Benjamin Šeško | 4 (2) | 173 | 6.80 | 2 | 0 | 1.12 | 0.00 | 2.6 | 0.0 | 4.7 | 2.1 |
All 11 competitive games, players with at least 150 minutes. Key passes are chances created for a teammate. Defensive actions per 90 describe a player's role and style, not how good the defending was.
Goalkeeper and defence
The goalkeeper is not the problem. Lammens has prevented about 2.0 goals more than the shots on target he faced would suggest (goals prevented, over nine games with shots on target). Darlow was 1.2 goals below that measure in the Cup tie.
Opponents' 129 shots were worth 16.8 xG. Most came from open play (87 shots, 11.1 xG, 0.13 per shot). Fast breaks were rarer (9 shots) but more dangerous at 0.25 xG per shot, and one of them was Lookman's goal at Atlético (0.88 xG). Of the 19 goals conceded, three came from set pieces (two at Hull, one at Leeds) and two on fast breaks (Brighton, Atlético).
Shot map of United's 194 shots, mostly central and inside the penalty area.
Shot map of the 129 shots United conceded, clustered inside and around the penalty area.
One substitution pattern stands out in the data. Against ten-man City, United used only 2 of their 5 substitutions (Šeško on 67 minutes and Zirkzee on 83) while level at 0-0 and then 0-1 down. The data cannot say why, and injuries or fitness may explain it.
Summary
Manchester United under Carrick are not collapsing, but the underlying numbers have turned negative. They have been behind on xG in all seven games since 13 September, creating 1.07 xG per game and conceding 1.64. They get the ball into dangerous areas (4th in the league for deep completions) but turn it into few clear chances, and when opponents press them or the game stays level late they concede good chances and late goals. Their own pressing has dropped from fifth most intense before Carrick to the middle of the league, and away from home they defend very deep. Selection has been unsettled (23 starters, seven centre-back combinations, eight players injured at Como) and about half of the individual xG comes from Mbeumo, Fernandes and Cunha. The goalkeeper is not the problem, and set pieces are a pattern to watch, not yet a proven weakness.
Last season's 2.29 points per game came with results well ahead of the chances (1.54 xPts per game) and is not a fair baseline. This season the results are about five points below what the chances imply, so part of the poor start is bad luck and some recovery in points is likely without any change in performance. A real improvement needs more than that: a steadier back line and midfield, better chance quality, and a consistent way of pressing at home and away. Eleven games is a small sample, so the trend since mid-September matters more than the total. The next three games (Bournemouth, Chelsea and Roma) will show it: if the xG difference stays negative while results improve, that is luck; if both stay poor, the problem is structural.
Press context, not checked against our data: beIN SPORTS on the poor start and the Brighton Cup tie, Gulf News on the fixture run, and ARY News (AFP) on Carrick's permanent appointment after the third-place finish.