He was simply saying that some websites generate their Corsi Relative % using season-wide data. A site like Corsica.hockey, which is my go-to, generates their Corsi Relative % using only the games that player was in.
So, AMac's stats look slightly better on some sites because the team through much of the year was an average possession team at ~49%. AMac was at a 49% CF%, so that looks OK. But, given that since he was called up, the team performed at a 55% CF%, that 49% looks very poor. On
http://stats.hockeyanalysis.com/ (which was posted in this thread), his RelCF% is only -2%. That's not good either. But, if you just isolate AMac's stats since his February call-up, using only the games he was in and not those he wasn't, it comes out to an adjusted -7.6%. That's brutal and flies completely in the face of the "he was really good and helped get them into the playoffs" crowd. And given that Ghost was hovering around break even to +1% before they were paired together, it's why many want to get him the hell away from Ghost....or Provorov now.
Even if you want to ignore AMac's numbers on the Isles (where he was the 3rd worst possession d man in the league) and just point to his RelCF% as a Flyer, he's 26th worst in the league out of 173 qualified d men by minutes played. He isn't getting difficult usage either, like some d men below him, in that span. 2 seasons ago, he got the 2nd easiest difficulty minutes with the 2nd highest ozone starts. Again, in his time this year, 5th hardest minutes and 3rd highest ozone starts. He went from bad d man getting among league-worst smoked in tough minutes to bad d man getting slightly less smoked in easy minutes.