Actually, that’s really easy.
Not every card has equal power in every deck. A card can be very strong in one deck, and middling in others.
Cards can be strong with certain pilots, and weak by the average.
Once you start counting deck power into the mix, even with a skill component, you inherently allow for people to start choosing their opponents.
Even in your example of 3v3 teams swapping in noobs, that does drop the rating to the point where you start seeing a different group of opponents, thus changing your match up spread in a way you control.
So when you start talking about a bunch of players with relatively equal skill, say the top 1k players, now they just get mirror matches because of the system matching by deck power as well?
There are a ton of little things that have to be considered and measured and weighted before any system could come close to weeding out all of that kind of behavior.
Or, you can just base it on win rate, which does just as good of a job of it in the first place.