Aston Villa vs Manchester City
Our model puts a Manchester City win at 41.3% in this Premier League fixture — close to a three-way split. Expected goals run 1.31 to 1.49.
What this probability has been worth
A number is worth what it has been worth. Across 1,904 predictions published before kick-off and scored against the result afterwards, the calls we priced between 35% and 45% came in 38.7% of the time — 700 fixtures in that band. The model claimed 40.6%, so the two are within 1.8 points of each other. In this range the stated probability has meant roughly what it says.
Where we disagree with the market
Bookmakers price Manchester City at 1.71, which implies about 58.5%. Our model has it at 41.3% — 17.2 points lower. That is a wide gap, and it is worth being plain about what it means: it is not a claim that the price is wrong and we are right. It means two methods that usually agree have landed far apart on this one, and only the result will say which of them read it better.
Attack against defence
The single rating hides where the mismatch actually sits. Splitting it, Manchester City's attack should find room against Aston Villa's defence, by 192 points. That is the mechanism behind the expected-goals line rather than a separate opinion about it.
How often they shoot
Manchester City take 16.8 shots a match against an average of 11.9. Volume like that usually means territory rather than precision, and it is why a side can dominate a match and still not win it.
Behind the defence
75% of the shots on target Manchester City face do not end up in the net, against 67.3% for the sides we measure. That is a keeper and a defence having a good run of it — and a figure that tends to drift back toward the middle, which is worth remembering before reading too much into it.