Model table
Premier League xG & xPTS table
Every Premier League club ranked by the model’s expected goals and expected points over this season’s settled matches, next to what actually happened.
Manchester City has the best expected goal difference in the Premier League (+6.3) after 6 rated matches.
Manchester City has collected the most points above its expected points (+5.8); Manchester United the fewest (-5.4).
Rated matches only: fixtures the model priced before kick-off. Sorted by expected goal difference. Fixture and result data last updated Oct 17
| # | Team | P | xG | xGA | xGD | xPTS | Pts−xPTS |
|---|---|---|---|---|---|---|---|
| 1 |  Manchester City | 6 | 12.0 | 5.8 | +6.3 | 12.2 | +5.8 |
| 2 |  Manchester United | 6 | 12.2 | 6.9 | +5.3 | 11.4 | -5.4 |
| 3 |  Arsenal | 5 | 8.7 | 3.8 | +4.9 | 10.1 | +1.9 |
| 4 |  Brentford | 7 | 10.6 | 8.9 | +1.7 | 10.7 | -0.7 |
| 5 |  Aston Villa | 6 | 8.6 | 7.4 | +1.2 | 8.9 | -3.9 |
| 6 |  Liverpool | 7 | 11.0 | 10.1 | +0.9 | 10.2 | +1.8 |
| 7 |  Brighton & Hove Albion | 5 | 6.8 | 6.4 | +0.4 | 7.1 | +2.9 |
| 8 |  Nottingham Forest | 5 | 6.5 | 6.2 | +0.3 | 7.0 | -2.0 |
| 9 |  Newcastle United | 4 | 6.2 | 6.3 | -0.1 | 5.4 | -0.4 |
| 10 |  AFC Bournemouth | 6 | 8.9 | 9.3 | -0.4 | 8.1 | -5.1 |
| 11 |  Everton | 5 | 6.5 | 6.9 | -0.4 | 6.6 | -0.6 |
| 12 |  Leeds United | 6 | 7.3 | 7.8 | -0.5 | 7.7 | +1.3 |
| 13 |  Chelsea | 6 | 8.8 | 9.8 | -1.0 | 7.6 | +4.4 |
| 14 |  Fulham | 5 | 5.9 | 7.1 | -1.1 | 6.1 | -4.1 |
| 15 |  Sunderland | 5 | 4.7 | 7.1 | -2.4 | 5.2 | -1.2 |
| 16 |  Ipswich Town | 3 | 3.1 | 5.5 | -2.4 | 2.7 | +0.3 |
| 17 |  Coventry City | 3 | 3.4 | 6.0 | -2.5 | 2.9 | -2.9 |
| 18 |  Crystal Palace | 5 | 4.9 | 7.5 | -2.6 | 5.0 | +2.0 |
| 19 |  Tottenham Hotspur | 6 | 6.8 | 9.6 | -2.8 | 6.4 | -3.4 |
| 20 |  Hull City | 3 | 2.7 | 7.3 | -4.6 | 1.6 | +5.4 |
What is an xG table?
An xG table re-orders the league by chance quality instead of goals: expected goals for, expected goals against and their difference. It shows which clubs are over- or under-performing their play. A side high in the real table but low here is riding good finishing or a goalkeeper in form, and usually drifts back.
The figures here are the model’s pre-match expected goals — its attack and defence ratings, refitted after every round — not shot-based xG. They are stored before kick-off and never revised, so the table is a record of what the model expected, matched against what happened.
Expected points (xPTS) convert each match’s probabilities into points: win probability × 3 plus draw probability, summed over the club’s matches. Points minus xPTS is the luck column: positive means the club has collected more than its performances were worth.
xG, xPTS and the other terms in the glossary · How the model rates attack and defence