The model calculates your chances of winning, losing or drawing a match based on your expected goals and your expected goals against - basically considering the quantity and quality of chances you created and conceded. Then it awards you expected points based on this.
Solskjaer outperformed his xPTS as he was 8th in the table but 11th in xPTS. Rangnick is third in the table and third in xPTS .
A more detailed comparison:
Coach | rank | xPTS rank | matches | points | xPTS | xG | xG against |
Solskjaer | 8th | 11th | 12 | 17 | 15.56 | 18.69 | 21.08 |
Rangnick | 3rd | 3rd | 9 | 16 | 17.39 | 16.43 | 9.56 |
These stats suggest you're creating noticeably more chances and concede drastically viewer chances under Rangnick. They also indicate that you aren't as clinical under Rangnick (should've scored 16.43, really scored 13) and underperformed your xPTS.
It probably isn't that easy. Might be the case that the more intensive play under Rangnick means that your players are less concentrated when finishing. Still, I'd say you're on the right track and finally seem to constanty outperform your opponents in terms of chance quality x chance quantity. As a comparison: This model had you fourth for last season as well with 65 xPTS, comfortably behind City (83), Chelsea (77) and Liverpool (69).