Context indeed. Even with the easiest of schedules to work OGS in after Mourinho's falling out they were vastly overperforming -- and not in the 'get the most out of the team' overperforming. They were putting up results that weren't likely, and this regression was always on its way (made worse, as it seems like the players are giving up on OGS already, though that is entirely speculation).
I have no idea how that guy calculates xG, but just looking at the first 3 events (1 miss for each team, 1 goal for United), it's hilariously off.
The first United chance is a completely unmarked volley by Lukaku from about 7 yards out off a corner. Apparently that counted for around .1 xG
The first Palance chance counts for .3 xG despite obviously being a worse chance. He's at a much greater angle and the ball is coming from a much tougher direction and given how the play developed, De Gea had the post covered.
Then Lukaku's goal also seems to count for about .1 xG He's shooting from the top of the box near the edge of the D, and none of the defenders are blocking any of the net.
Then I'll jump forward to Palace's goal. Completely unmarked header from the top of the 6 yard box with the keeper out of position since it was a hard cross. Apparently only worth about .1 xG despite being the clearest chance of the game.
Young's goal is worth about .2 xG, he was in acres of space running into the box and was able to run to the edge of the 6 yard box and crush a shot with basically no pressure. A professional footballer should score more often than not in that position.
Those are just 5 examples I saw quickly and I don't see how they're ranked like they are.
xG is a pretty garbage stat right now, at least the public versions.
Using understat.com as a reference since it's easy and public: In the last 5 years, only 4 times has a big 6 side underperformed their xG stat, Arsenal in 15/16, United in 16/17, and both Liverpool and City slightly this year. Over that timeframe the average a big 6 side has out performed their xG stat is by 5.5 goals.
Seems pretty clear to me that something is missing if the best teams frequently outperform an expectation.
League wide this is the first year that teams have outperformed their xG stat, and by a total of 16 goals. The past 4 years, the PL has cumulatively underperformed it's xG stat (or outperformed it's xGA stat) by 160 goals cumulatively. Seems more likely the stat is poorly constructed than that defenses were getting lucky for 4 straight years league wide (in one year by a whopping 80 goals)