Another radical idea on how to fix dynamic queue
One of the biggest complaints about DQ is how it gives premade teams an unfair advantage. From that data we're seeing winrates getting higher and higher the larger your party gets.
The easiest way to fix this is to change the MMR system to not be a 50/50 based system. Instead make it so it's an average off of the average that team should be winning at. Meaning if you have 5 players of all equal skill levels and all in a team they should have something close to a 60/40 split while a solo team that go against that should be valued at a 40/60 split. If it's two 5 mans going at eachother it should obviously revert back to a 50/50 split.
A few things will need to change in order to make players understand why they're getting larger/smaller LP/MMR drops from game-game. First it needs to be made clear what type of game is being played. Meaning if it's a 5 man party vs 5 solo or 5 man vs 3/2 parties. This'll help players understand why they lost 20 LP one game and only 10 for the next. Second it needs to be communicated that players who match together are expected to win more in the client. Finally it'll need to be explained that solo players should not expect high winrates (60%+) while climbing, and they shouldn't be dissuaded even if they aren't winning 50% of their games but they're still climbing. Solve those three issues though and dynamic queue should work as an accurate measure for player skill.