Alright I'm at lunch typing this up from memory since I'm not at a computer, I'll probably follow up with exact numbers later.
This time instead of a simple goal seek I used as evolutionary approach.
I calculated win percentages by aggregating the matches from the last 3 vintage challenges and then recalculated win percentages as if all 3 were a single large event. I excluded decks that didn't get 6 instances across three events, so that meant no delver eldrazi or non outcome combo. I left big blue and blue control as separate decks rather than combining them into a drain pillar (may do that tonight).
I started with each deck shops, dredge, outcome, oath, mentor, big blue, blue control, and other having equal share of the meta. Then based on the win percentages and meta share I calculated each deck's win percentage against the field. For each deck with a below 50% win percentage, I assumed 10% of the people playing that deck would move to a deck that had a >50% chance against the field with the shares being proportional to how much the new deck was beating the field.
I then repeated this 2000 times to see if a stable meta emerged with a balance of decks that all won about 50% against the final field.
That's didn't happen, instead the meta settled into an oscillating patter that repeated every 150 iterations or so.
In this repeating pattern oath and outcome are unplayed. They get forced out of the meta pretty quickly (less than 50 iterations) and never come back.
The two main pillars are shops and big blue. Shops goes from about 20% to about 57% and back. Big blue goes from around 25% to around 60% and back. These two decks together are never less than 70% of the meta and sometimes make up as much as 80%.
The mix of the remaining 20%-25% drives where the main two decks are in the cycle and which of them are waxing or waning.
Of these, other goes from 0% to 5%, blue control goes from around 5% to 15%, and mentor also is in the 5% to 15% range.
Obviously all this analysis depends on the win percentages being accurate, and we don't actually have the sample size to derive accurate percentage information. So don't take this too seriously. But it gives an interesting picture of how things play out if what we are seeing is actually accurate.
Also more people need to be playing mana drain.