Five games in, and the international break gives these numbers a fortnight to sit before anyone can do anything about them. That makes this the most useful week of the season so far to look at the gap between what Premier League teams have earned and what they have got.
Expected goals weight every chance by how often a shot of that type is scored. Expected points run those chances through the match to ask what a league table would look like if the finishing had been average at both ends. The table below is the real one; the gaps are what is not yet priced in.
AFC Bournemouth have three points from a possible fifteen. On expected points they are on 7.9, which is the fourth-best figure in the division and would have them in the top six. A gap of -4.9 is the widest in the Premier League in either direction.
Their Gameweek 5 was the pattern in miniature: beaten 1-0 at home by Liverpool on 1.13 expected goals to 1.55 - a close game they lost. They have conceded eight goals from 6.1 expected, and scored six from 7.6. Both ends are running against them at once, which is the least sustainable way to be bad.
At the other end, Manchester City have fifteen points from 9.7 expected, +5.3. Thirteen goals from ten expected. Their 5-3 win over Sunderland is the single clearest example on the card: eight goals from a combined 5.36 expected, and Sunderland actually created the better of it at 2.79 to 2.57.
Manchester United have five points. They also have 10.4 expected goals, the second-highest total in the division behind Brighton, and 9.1 expected points against the five they have collected.
Their 1-1 draw at Fulham followed the script: 2.08 expected goals to Fulham's 1.50, one goal to show for it. The defensive number is the one that should worry them more - 5.2 expected goals against, third-best in the league, and eight actually conceded. A side creating the second-most and conceding the third-fewest chances does not usually sit fourteenth.
Fulham are in the same territory from the other side of that match: two points against 6.5 expected, a gap of -4.5, and flagged in this column a fortnight ago at exactly the same figure. Nothing has corrected.
All ten fixtures have expected goals in. Two scorelines are doing a lot of lying.
| Match | Score | Home xG | Away xG |
|---|---|---|---|
| Brentford v Chelsea | 3-0 | 2.13 | 0.89 |
| Tottenham Hotspur v Aston Villa | 2-3 | 1.45 | 1.53 |
| Everton v Ipswich Town | 1-0 | 1.59 | 1.29 |
| Newcastle United v Hull City | 2-1 | 1.43 | 1.74 |
| Brighton & Hove Albion v Arsenal | 3-0 | 1.60 | 1.48 |
| Nottingham Forest v Coventry City | 0-1 | 1.41 | 0.56 |
| Manchester City v Sunderland | 5-3 | 2.57 | 2.79 |
| AFC Bournemouth v Liverpool | 0-1 | 1.13 | 1.55 |
| Leeds United v Crystal Palace | 0-0 | 1.42 | 1.15 |
| Fulham v Manchester United | 1-1 | 1.50 | 2.08 |
Brighton 3-0 Arsenal is the one to look at twice. On chances it was 1.60 to 1.48 - about as even as a match gets - and it finished three-nil. Arsenal are still +2.1 on expected points for the season, so the result cost them more than the performance did.
Nottingham Forest 0-1 Coventry City runs the other way: Forest created 1.41 to Coventry's 0.56, two and a half times as much, and lost at home. Nottingham Forest are -3.1 on the season with the fourth-best defensive expected-goals number in the league.
Coventry City have scored one goal from 4.2 expected. That is the worst finishing return in the division by a distance and it is the reason they are nineteenth rather than mid-table. Three points from 4.3 expected is a smaller gap than the goal numbers suggest, because they have been close in games they lost.
Crystal Palace have conceded 11.3 expected goals in five games, the worst defensive number in the Premier League and more than double Arsenal's 4.6. They have eleven actual goals against, so this one is not bad luck - it is a defence giving up chances at a rate nobody else is close to.
Brighton have sixteen goals from eleven expected. Most goals in the league, best expected-goals total in the league, and five goals of finishing on top. Only the last part of that is unlikely to hold.
Three teams are priced on a league position their performances do not support, and the break means nothing will change for a fortnight.
AFC Bournemouth at -4.9, Fulham at -4.5 and Manchester United at -4.1 are all creating and conceding like mid-table or better sides while sitting in the bottom six. Newcastle United at +3.8 and Manchester City at +5.3 are the other way round. Our own testing across 1,245 fixtures says the tradeable half of that is fading the over-performers - sides running more than half a goal a game ahead of their chances then win 3.2 percentage points less often than the model rates them, while the under-performers beat their rating by only 0.8, which is noise.
So the read going into Gameweek 6 is not to back Bournemouth and Fulham to bounce. It is to be careful about what Newcastle United and Manchester City are priced at.
18+. Please gamble responsibly - begambleaware.org. Expected goals data via our own import, updated as each round settles.