Which Premier League positions are real?
Brighton's goal difference is a fluke, but 3rd place mostly holds. Forest, Man United and Sunderland are better than their places.
Forest, Man United and Sunderland have far fewer points than their chances
All 20 teams as a table
| Team | Pts | GD | xG diff | GD minus xG diff | Proj. pts |
|---|---|---|---|---|---|
 Arsenal | 24 | +9 | +7.3 | +1.7 | 82 |
 Man City | 21 | +10 | +7.7 | +2.3 | 77 |
 Brighton | 19 | +17 | +3.8 | +13.2 | 69 |
 Chelsea | 16 | +4 | +0.9 | +3.1 | 60 |
 Liverpool | 15 | +2 | +3.0 | -1.0 | 60 |
 Brentford | 14 | +7 | +2.1 | +4.9 | 58 |
 Bournemouth | 12 | +1 | +4.9 | -3.9 | 57 |
 Tottenham | 12 | 0 | +1.6 | -1.6 | 53 |
 Aston Villa | 12 | -3 | -5.9 | +2.9 | 45 |
 Hull City | 11 | +1 | -4.4 | +5.4 | 45 |
 Leeds | 11 | +1 | -0.9 | +1.9 | 49 |
 Newcastle | 11 | -2 | -5.3 | +3.3 | 46 |
 Ipswich | 11 | -4 | -5.3 | +1.3 | 44 |
 Everton | 10 | -1 | -2.7 | +1.7 | 47 |
 Fulham | 9 | -2 | +3.6 | -5.6 | 50 |
 Coventry | 9 | -7 | -6.1 | -0.9 | 39 |
 Man United | 7 | -5 | +2.1 | -7.1 | 45 |
 Crystal Palace | 7 | -12 | -10.3 | -1.7 | 31 |
 Nottm Forest | 6 | -6 | +4.5 | -10.5 | 45 |
 Sunderland | 5 | -10 | -0.8 | -9.2 | 38 |
Nottingham Forest have the Premier League's 4th-best expected-goals difference and 6 points. Expected goals (xG) measure how many goals a team's chances would produce for an average finisher. Manchester United (9th, 7 points) and Sunderland (12th, 5 points) are in the same trap. All three sit in the bottom four more because of finishing and goalkeeping than how they play.
Brighton are the mirror image: +17 goal difference from +3.8 in xG. Yet their points are only about 4 ahead of their chances: most of the extra goals went into 5-0 and 4-0 wins.
Brighton are 13 goals ahead of their chances; Forest are 10 behind
Subtract each team's expected-goals difference from its goal difference. What's left is finishing, goalkeeping and luck.
Brighton's swing is more than double anyone else's
BrightonHull CityBrentfordNewcastleChelseaAston VillaMan CityLeedsEvertonArsenalIpswichCoventryLiverpoolTottenhamCrystal PalaceBournemouthFulhamMan UnitedSunderlandNottm Forest- Brighton (+13.2): 24 scored from 17.0 xG, 7 conceded from 13.2. Of 192 big-five league team seasons in 2024-25 and 2025-26, only Bayern Munich's 2025-26 start (+13.6) was bigger.
- Nottingham Forest (-10.5): 6 scored from 14.6 xG. No team in that sample had a +4.5 xG difference and 6 points or fewer.
- Sunderland (-9.2) and Man United (-7.1) are next, with far less attention.
Brighton's lead built in blowouts; Forest's hole built every week
Brighton gained most of their 13 goals in three games. Forest fell short in 8 of 9.
Brighton's swing came in jumps; Forest's slid almost every week
- Brighton: 5-0 at Coventry on 2.5 xG, 3-0 over Arsenal on 1.5, 5-0 over Palace on 2.4. On Saturday they won 1-0 at City, out-created 2.0 to 1.0.
- Forest: a positive xG difference in 7 of 9 games, one win, and no goals in 4 home games.
Goals above or below xG almost fully reset after 9 games
We split 192 big-five league team seasons (2024-25 and 2025-26) into their first 9 games and the rest. Two seasons is a small base, so the exact figures are rough; the pattern is not close.
Big early swings vanished after game 9, in both directions
- The luck doesn't carry: goals minus xG over the first 9 games had a 0.08 correlation with the rest of the season. xG difference carried at 0.75.
- Points follow: hot starters fell from 1.78 points a game to 1.44; cold starters rose from 0.75 to 1.10.
- Goalkeeping resets too (-0.09). Only shot accuracy carried a little (0.20).
Team by team: the players behind the numbers
- The 13-goal swing: Verbruggen's saves +4.5, opponents missing the target +2.8, Brighton's finishing on target +2.6, own goals +2.0, shot placement +1.4.
- Scoring is spread out: Charalampos Kostoulas, Pascal Groß, Yasin Ayari and Jack Hinshelwood have 3 each, none more than 1.1 above his xG.
- Morgan Gibbs-White's only goal is a penalty, from 2.3 xG. Dan Ndoye has 2 from 2.0 with 5 big chances missed.
- The caveat: their shots on target are worth 7.7 expected goals on target, half their xG, and shot accuracy is the one piece that carries over a little.
- Striker Benjamin Sesko has been out since Sept. 16 (shin); Manuel Ugarte and Patrick Dorgu are also out, and Leny Yoro is suspended.
- Their xG difference is about even (-0.8), 10th after schedule: mid-table numbers.
- Brian Brobbey (3 goals, 3.3 xG), Dan Ballard and Romaine Mundle are listed out. Two of their 7 goals are Enzo Le Fée penalties.
Penalties and the schedule don't explain the gaps
After schedule, Forest, Chelsea and Man United sit 6th to 8th
ArsenalMan CityLiverpoolBournemouthBrightonNottm ForestChelseaMan UnitedBrentfordSunderlandFulhamLeedsTottenhamEvertonAston VillaCoventryIpswichHull CityNewcastleCrystal Palace- Penalties: only 11 scored all season, no team more than 2.
- Schedule: Forest have had an easier run, so they drop from 4th to 6th. Chelsea and Sunderland rise.
Everyone sees Forest. Fewer see Sunderland.
- ESPN's Bill Connelly (Oct. 29) files Forest under "good, but unlucky": "The fundamentals are in great shape." He puts Sunderland in "bad, but unlucky."
- Forest manager Oliver Glasner, via Futbol Chronicle on Yahoo Sports (Oct. 23): "It feels unfair," adding: "We are the only ones who can change it."
What the data adds: Brighton's goal difference will shrink, but 3rd place is not the fluke. The bigger mismatch is Sunderland: mid-table chances, last-place record.
Forest and United project to the lower half, not the drop; Brighton stay 3rd
We projected the last 29 games from points and xG difference per game, fit on the 192 past team seasons, then simulated the table. The odds are rough (two seasons of history, no fixture list), so we give ranges.
Forest and United project about 45 points; Brighton hold 3rd
Arsenal24 pts now, 1stMan City21 pts now, 2ndBrighton19 pts now, 3rdLiverpool15 pts now, 5thChelsea16 pts now, 4thNottm Forest6 pts now, 19thMan United7 pts now, 17thSunderland5 pts now, 20th- Forest project about 45 points, in a pack of six teams at 44 to 46. Chance of going down: roughly 15 to 20%.
- Man United also project about 45 (roughly 20%). Sunderland project 38 (roughly 40 to 50%), alongside Crystal Palace and Coventry.
- Top four: Arsenal and Man City 85% or better, Brighton about 60 to 70%, Liverpool and Chelsea about 30% each.
Questions people ask
Are Brighton for real in 2026-27?
Their position mostly is; their goal difference isn't. Brighton's 13-goal swing over their xG won't repeat, but they rank 5th in xG difference, sit only about 4 points above their chances and project 3rd.
Why are Nottingham Forest in the relegation zone?
Finishing. Forest have the 4th-best xG difference (+4.5) but 6 goals from 14.6 xG, none at home, and the league's lowest share of shots on target (20%).
Will Manchester United be relegated?
Probably not. United are 17th but 9th in xG difference. Snapp AI projects about 45 points and roughly a 1-in-5 chance of going down, a rough estimate.
Does outscoring your xG last in soccer?
Mostly no. In 192 big-five league team seasons, goals minus xG over the first 9 games had a 0.08 correlation with the rest of the season, while xG difference carried at 0.75.
How we did this: Snapp's soccer warehouse (Sportmonks xG and expected goals on target), Premier League matches through Nov. 1, 2026 (89 of 90 matchweek 1-9 games; Newcastle vs Everton on Nov. 2 not included). xG difference = team xG minus opponent xG. Goal difference minus xG difference counts own goals as goals; own goals carry no xG. Goalkeeper figures compare expected goals on target faced with non-own-goal goals allowed. Carryover tests and the projection use 192 team seasons from the Premier League, La Liga, Bundesliga, Serie A and Ligue 1 in 2024-25 and 2025-26 (first 9 games vs the rest). Projection: rest-of-season points per game = 1.007 + 0.269 x points per game + 0.343 x xG difference per game (R-squared 0.58; typical miss 0.30 points a game, about 9 points over 29 games). Probabilities come from a 200,000-run simulation that treats teams as independent and ignores fixtures, so read them as rough; ranges cover the base run, a 30% wider miss and projections shrunk 20% toward average. Only two seasons of xG history exist per league, so correlations carry real uncertainty (about +/-0.14 at 95% for the 0.08 figure). Schedule adjustment adds the average xG difference of each team's opponents. Injury lists as of Nov. 1. xG models differ by provider: the BBC had City-Brighton at 1.92 to 1.00, Sportmonks at 2.03 to 0.98.