I have no clue what XG is, except what you have told me right now.
So, I won't analyze those stats since I don't know how they work, what are the limitations and towards which type of matches/teams/number of shots is that system biased.
I guess that XG, like all stats has some limitations and things which you don't see on stats.
XG is the best metric we have so far to determine how good a chance (a final attempt taken) created by a team is.
This is done my measuring a number of factors, like distance from goal, position of the GK, players in front of the ball, speed of the ball before taking the attempt, angle etc
For example a shot taken with an xG of 0.01 is not the best attempt to score a goal: it goes in only 1 out of a 100 times. On the contrary a shot taken with an xG of 0.3 is a quite good chance: it goes in 3 out of 10 times. To get an idea of how the percentage goes: a PK has an xG of 0.75
Now, if you see the xG score line of a game (which is the outcome of adding up all the attempts, excluding penalties, and their respective xG counts to determine a scoreline for both sides) you get a sense of how 'well' both teams performed, especially if you combine it with an xG shot map. 'Well' means here how effective (irrespective of playing style).
Low xG means inability to create danger, inability to translate possession into meaningful chances, while high xG means the contrary.
Discrepancies between an xG scoreline and the actual scoreline happen for three main reasons:
i) Luck (big part in football)
ii) Referee decisions (a penalty gives a very probable goal to a team with 0.75 chance)
iii) Individual over/under-performance on both ends
Regarding iii, for example Liverpool's underlying numbers at the end of 17-18 season were not far from their underlying numbers of 18-19, but the inclusion of Alisson in the place of Karius made a significant change. They managed to get scorelines that reflected their defensive performance better, because the GK was not conceding chances with low xG.
Likewise, having Messi upfront gives a team the luxury to turn frequently attempts of low xG (for example free kicks from outside the box with 0.05 and below) into goals.
A good team offensively = a team that does not rely on individuals to turn low xG attempts into goals, but provides them the opportunity to finish off high xG attempts
A good team defensively = a team that does not give away significant chances to the opposition (attempts of decent and high xG) relying on the GK to save them.
If you calculate points gained after every fixture based on xG scoreline and not the actual scoreline, you can create a table of 'what it would have been had the actual scorelines resembled the effectiveness of the team'
There you see that EV record is pretty horrible. There are many sites/resources you can look at, i will give you one: According to understat:
The 17/18 xG tally of Barca was 79.44 points, 13 and more less than the actual 93. While Real Madrid's was almost the same 79.99, 4 more than the actual 76 they got
The 18/19 xG tally of Barca was 73.96 points, 13 again less that the actual 87.
So, we see that even his best 17/18 season was based largely on Messi over-performing upfront to turn insignificant chances into goals, and Ter Stegen over-performing at the back to save the team's ass from malfunctioning defensive reactions.