I’ve talked myself into it. Despite the problems, I think matchmaking should be based on Elo ratings, like in chess tournaments. Your rating goes up with a win, down with a loss, and you are grouped with players who have similar ratings.
Within a rating group, differences in rewards between winners and losers should be relatively modest; but reward differences between different rating groups should be substantial. That way competetive players have a reason to keep improving their rosters and ratings, but more casual/collector players continue to get relatively fair matchups, but settle for lower rewards in general.
I feel like this would solve almost every problem with the pairings in the long run. There would be no need for CG to crunch numbers on individual rosters to figure out how much each zeta or G12 is worth. Your rating just emerges naturally out of your actual performance in prior events.