Stat of the Match: Manchester United 0-3 Liverpool
Erik ten Hag saw the xG. It wasn't that high.
“I’ll be honest, I think Liverpool did brilliant. How they took over the turnover and how they finished was so good, but I saw the xG, it wasn’t that high.”
Erik ten Hag to Sky Sports after the game
Arne Slot continued to make big team management look easy as he led Liverpool to a 3-0 victory over Manchester United at Old Trafford. The other two Premier League seasons which began with the Reds winning three games to nil saw them finish second. If not a position to make your ultimate aspiration, it shows what is possible with a perfect start.
It was a performance which hinged on the hour mark; solid and incisive before, a little open and wasteful thereafter. Liverpool could afford for their level to drop because by that point they held a three goal lead. When Mohamed Salah gave them that advantage, it was only his side’s fifth shot of the game.
The quintet of chances were collectively worth less than one expected goal (see below). Was Erik ten Hag therefore correct when he said that the xG “wasn’t that high”?
Like any data, xG in small samples doesn’t tell you much. It can frequently mislead for a match, never mind an individual shot. More important than their mathematical value was that the three chances Liverpool converted were classified by Opta as big, meaning it was reasonable to expect the attacker to score.
(There’s a lot of suspicion and derision among the analytics community regarding big chances, with their subjective definition. But trying picturing a chance valued at, say, 0.24 xG and another you’d call ‘big’, and see which is easier to call to mind).
As recently investigated on here, the Liverpool of 2023/24 would at times build a large expected goal total in part through a heavy accumulation of low value chances. Analytics legend Mark Taylor proved over a decade ago that it is better for a team to have fewer but higher quality goal scoring opportunities than amassing the same expected goal tally from more shots. The Reds did the former at Old Trafford on Sunday.
Ten Hag made no reference to the post-shot expected goals (PSxG) with his comments, but the data is worth a look as it helps explain the outcome better than the raw xG does in this instance.
None of Liverpool’s three goals came from chances valued at much better than a one-in-three likelihood of a goal. Drink in those figures in the post-shot column though. They add up to 2.27, meaning Liverpool had around a 40 per cent probability of scoring three goals (as represented by blue in the below chart) once the finishes had been applied to those chances, rather than the two per cent possibility the basic expected goals suggested (in red).
The xG of a shot doesn’t have to be that high if a chance is clear-cut and the placement is spot on. Offering over half of the net to a player who has outperformed his league expected goals by nine on his stronger foot isn’t the smartest move either.
It would be easy to take some easy shots at the United boss here. Scrutinising anything a manager says to the media is a pointless exercise though. They have a message to project to the watching world, their players and themselves. His comments belied something of a misunderstanding of what the metric can tell you, though, and it’s something which permeates many pundits and supporters alike.
Another major problem with football statistics as they are so often presented in the media is that they are offered without context. The Reds’ non-penalty xG figure of 1.7 doesn’t sound that high, but it isn’t that straight forward either.
Manchester City have conceded at least that total in only 17 of their last 269 Premier League games and just twice at home since the summer of 2017. While those figures are higher for Liverpool, at 35 and five respectively, by the standards for which United are presumably aiming 1.7 xG actually is quite high.
The Red Devils’ manager might want to reflect on the fact that his side have allowed this amount or more 26 times in a little over two years. In terms of frequency of occurrence, that sounds pretty high to me.
Really good Andrew 👍
Very good detail. Hopefully Mo bombshell on Sky will get Hughes off his arse