Day 2 of.. oh right I should have mentioned this last time.
The other solution that some of my Friends and I have used over the years. We do what we call Scrims. Get a group of 10 friends online and play only with and vs ourselves in custom games. As most of them are folks I have known from Clans in other games or IRL hommies we all have contact on social media. So we made a FB group for Lol and would set up times and dates in advanced. Some times if we are low we just shoot out invites to random LOL friends listers to join in. As long as you don't have some one so good they need to be handy capped cuz their team just wins cuz they are on it. (fucking masters buddies just demolishing games). Some times the handicap is works out fine cuz we run a 4 v 5 game.
Anyhow been thinking about your issue. I totally can be wrong but here is my current guesswork on it.
Unlike Elo, most MMR systems are not Zero sum games. Side effect of adapting from 1v1 to 5v5, part of accounting to track each player individually is you cant give everybody in the match the same stakes. All it takes is some uncertainty variance and having the MMR gains and Losses balance out is impossible. Actually, that's not entirely true, it can be done but its honestly unfair to some unlucky blokes. So I wouldn't run a Zero sum team MMR but I cant accurately promise you Riot wont. They shouldn't thou... no good reason for it and they as a company tend to care far to much about player perception to leave something like that in.
Normal MMR does not get reset ever far as I know, so the extra MMR build up is gonna do things to the MM. Accounts can get inflated depending on how many players are around. When its not Zero sum, MMR can be added to the systems total. Soft resets operate as a yearly place to remove some of the excess MMR to prevent it bloating accounts and the system averages.
https://www.leagueofgraphs.com/rankings/rank-distribution
Bronze to Gold is listed at 82.54% of the games population.
For what I know the MM in normal has to make a game. There is no documented case (I have ever seen) of it telling players theres just no one on in a MM thats fair, go play a different Que. Because it seems to HAVE to make a match then I think your at a higher risk to get dicked over when it needs a team for other 5 mans above you. I think that the system has to look down dam near 100% of the time it cant find a match for the players representing under 14% of the games players with similar ranges. I think expecting it looking up is completely hopeless as that's going to be 1.4% of the games population. Part of this is cuz those mismatches where you guys get really dicked are all constantly high MMR. I think by mainly playing normal your gonna see bad matches more then if you guys had been in a que with higher participation and tighter MM ranges. When looking at High MMR accounts in normal I think the safe assumptions are that there is going to always be less of them online. My thought process here is each player accounts for a higher % of that tier and that the overall odds of them queuing in normals is lower if only because theres less of them to que.
I might retract this opinion in the future, I am on the fence and trying to figure out exactly why your getting the experience you do. I would love to get some more Data. How long do you guys tend to spend in que? How many players do you usually que with? What's the average MMR for your fair games? Whats the MMR look like when you solo normal, for several games? How many times is it the high MMR stomp teams are qued togeather?
You don't need to answer this because it will still just be 100% guesswork on my end.
I like puzzles and I like learning. So your post has been pretty interesting for me.
So basically My logic is Filling in games when players are in a standard range is a lot easier because achieving a smaller range of MM is easier to do for a range that represents 80% compared to one under 15%. I would not bet $ that im right or even accurate on this thou, I really do want you treat it with some suspicion.
