Cool, sounds like we are very close @Jainus. I guess the fundamental difference I was noting is that you indicated a seemingly top-heavy weighting for your matchups, while I’m suggesting a more random/open matchup across the group. But I thin our concerns and goals sound the same, just different ways to achieve them 
I agree, was just using it more to illustrate that the model could be easily (and significantly) tweaked just by changing that one variable (group size). Looking at the actual data available (Top 50 Guild League Table History , from @actreal), I’d make the group size 30, but that’s only based on that incomplete data (there are likely a bunch of very active guilds that simply haven’t been around long enough to make the top 50 yet).
But again, I’d be interested in hearing the official view 