I'm still on the fence on how much ranking comes into play.
When we did the GW I noticed that at every step of the way there was as much as a 4 win differential between the teams around us.
That implies that the even though you're in the top 300, you don't actually have 300-400 other guilds to match with. Especially if you happen to have more wins than everyone else around you.
On the other hand, if it was strictly by rank then there's the possibility of spenders running into non-spendings more often, which would lead to spenders being able to run up the score to climb ranking (or at least keep pace easier).
That would be unfair to people trying to rank, so that would be a pretty good reason for SS to tie ranking into match making just to avoid giving randomly skewing the rank race.
Now taking personal experience, we were consistently in the top-end of wins for our rank (top-200) and only had the score run up on us one time (don't know what rank that guild was). So it seems unlikely we dodged all the heavy spenders despite what I said at the start (although I know we faced at least two ~50 ranked guilds, going 1-1 and the team we lost to *did not* run the score up so maybe running the score isn't that common).
I guess I'm leaning more towards subdividing by wins and rank after typing that out, I'm curious to see what numbers you come up with Spacebug.