The short answer is I’ve found you have to look at pathways to victory, and how those are accomplished. Is it raw card combos, is it an interaction of actions and cards, etc. “How would one string together points to make a victory, and how many different pathways are there to do that?” As part of doing this, you may find that you have an attribute (e.g. magic, or melee) that features more prominently in the first pass of pathways and you adjust those accordingly. Nothing is going to save you from playtesting though, there will have to be some brute force effort thrown at it to verify assumptions.

The biggest issues with card combo games is solving the deck size problem that exists (see Ark Nova, Pax Porfiriana, etc for games that exhibit the “big deck problem”). The smaller the deck, the more it becomes “how does the limited pool of cards get combined and associated with a player” in a brittle fashion vs at larger deck sizes, you get “will the cards I need show up in the game at all.” There are mechanical ways of reducing blind spots associated with large and small decks, but understanding the problem is a helpful starting point.