Every Monday the results table lies a little, and expected goals says by how much. Eight of Gameweek 3's ten games have full xG in our data as of Monday afternoon (the two Sunday London games land later in the import cycle - we will cover them next week), and the eight tell some strong stories. All numbers are from the Statz match data, at the time of writing; the full running table lives on our Premier League xG table.
Fulham against Crystal Palace produced the widest gap between performance and result this season has seen: Fulham 3.89 xG, Palace 1.04 - and Palace won 3-2. A team creating nearly four goals of chances loses that game roughly once in a dozen tries. None of that pays out anything for Fulham, but it is exactly the kind of defeat that leaves a side underpriced the following week, and exactly the kind of win that flatters the team that got it. The model will not be moving much on either.
Our Saturday acca backed Brentford at home to Sunderland and it finished 1-1 - and the xG says the pick, not just the result, was wrong: Sunderland 2.40, Brentford 1.26 at the Gtech. The promoted side created the better chances away from home for the second time already this season. Their prices still carry a promoted-club premium; their numbers increasingly do not justify it.
Same flavour at St James' Park, where Bournemouth's 1.36 to Newcastle United's 0.76 made the 2-2 a fair result at worst for the visitors.
Hull City 0-0 Aston Villa extended the season's strangest split: Villa sit in the relegation places, and their underlying numbers keep saying they should not. They out-created a Hull side who started the day third, 1.34 to 0.71, and still could not score. The model backed Villa at 9/10 on Saturday and lost; the xG says it would make the same call again. Villa start their Champions League campaign in Bruges on Tuesday - the match preview has the numbers.
Honesty cuts both ways. This column's standing position has been that Liverpool's underlying numbers were better than their results - and on Friday they won 2-0 at Ipswich while producing just 0.71 xG to the home side's 1.03. That is the reverse of the story we have been telling, and one game does not settle the argument either way. What it does mean: do not chase Liverpool's prices down on the back of two wins. The numbers have not turned yet; only the results have.
Everton 2-2 Manchester United and Arsenal 2-1 Chelsea are awaiting their xG import and will lead next week's review.
Three positions, all model-flavoured rather than promises: Fulham are the buy - the market punishes the scoreline, the numbers say the performance was elite. Sunderland away are no longer a fade - twice now the travelling xG has been genuinely good. And treat Palace's win and Liverpool's numbers with the same scepticism - both results outran their chances. The projections behind all of this feed the Daily Acca and Bet Builder Tips every morning.
xG figures are from the Statz match data import and can receive small provider corrections after publication. 18+. Please gamble responsibly - begambleaware.org.