At the beginning of this thread, when I looked at offensive production from 2017/2018-present, you came in and shrank that sample size by a year, because that made Panarin look better.
Now, when looking at defensive impact using a stat in a highly problematic way, you ignore this current season in which Matthews has been better than Panarin by your metric, and put emphasis on the previous 3 years, because that makes Panarin look better.
Through this, you pretend that playing with mediocre relative linemates, largely a rotating cast of rookies, in front of the likes of Gardiner/Zaitsev is in any way comparable for defensive help to playing with better linemates, largely established players or stars, in front of the likes of Keith and Jones.
Maybe at ES, depending on what samples you use, how you value primary points, and how sustainable you think Panarin's current level is (hint: not very). On the PP, Matthews is better.
Panarin is at 2.88 xGA this season, significantly worse than any other year, largely because xGA tends not to transfer much with a player, and is highly dependent on external factors to the player in question, mostly defensemen.
Matthews is at 2.61 xGA (not sure how that's "even"), which is in line with his 1st year in the league, not a complete 180.
So Matthews' improved metrics don't count, because you apparently think that as players improve with age, experience, comfort in the league, and enter their prime, they get worse defensively? Maybe that should give you an indication that you are measuring defensive impact in highly flawed and problematic ways.