First up, https://na.op.gg/summoner/userName=Puffin+Chucker+ I don't see any plat or diamond player in your visible history.
Not much can be done if one of them is a smurf because the only thing reflected on an account is that accounts personal history.
Real talk my man, you probably will not get a rioter to comment. I like nerdy things like game designe and systems behind things. So unlike a lot of posters I have taken some time to study and learn about MMR systems. I am not at all an expert on the topic... if i was id work in the industry ^.^but I feel like I am gonna be more helpful then people who just make up their opinion on the topic with out researching it. So here comes a Wall.
You need to keep in mind that the entire Tier system has (almost) nothing to do with MM. Its not used by the system to build games. I say almost because The way Riot built "Duo Queing" creates issues. Its restriction for ranges are off tiers not MMR. How its implemented, It wont break the game or ladder, like you said when people get outclassed they get stomped and its not fun. Duoing does make problems from a strict Quality standpoint but as its not breaking the system and the player base really enjoy it so Riot stands to lose more player satisfaction if they change it.The simple fix is to make duoing off the same MMR range MM considers Fair to pull from, and also would probally need some tweaks for the Sections of the ladder that represent player populations under 1% of the games base.
Now that the exception is out of the way, a massive majority of the time the actual tiers don't matter and its silly people think they are instantly a good marker for accuracy on any 1 players skill. 10 placement games gives you a tier. Your going to see people go on about win rates and win streaks but most the time they don't know wtf they are talking about.
MMR is a balance. its not starting at 0 either. Every new account in a season has the same balance. The reason its not 0 is because you want players to be able to go down on the ladder at get put into lower skill games any time they struggle at the starting value. If you cant move them down, they stay in that range for MM and basically are expected to just always struggle.
People who don't know the system make a big deal about win streaks or win rates. I'm going to try and help you see win% and "streaks" impact in action and what your going to actually want to look for, Consistency over time is a huge deal.
Lets make some numbers up and show some principles behind the system. Lets call the starting value 1400 points. The Adjustment formula for MMR systems is some derivative/modification of Arpad Elo's Rating system. He used Rn= Ro + K(W-We) Alone that kinda looks like gibberish, So here is what those variables represent.
_RN _is your new rating at the end of a game.
_RO _is your rating befor the game was played.
_K _is used to represent uncertainty (if you meet any one who talks about MMR and they don't know what a K factor is that person has not done even a single google search on the topic and should not be trusted as an accurate source.)
_W _is the result of a match. Win, loss, draw. In league we only have win or loss.
_We _is what the system thinks the result of a match would be. Abbreviated, you can say that this is created by looking at Ro of players.
So in action, im going to use the number I made up early, 1400 for RO. So Rn = 1400 +36(1-0) RN= 1436. Now when you play your next game your R0 will be 1436 and your RN is going to change off the result of the next game, win or loss. The numbers im using for K I picked arbitrarily**, K factor is a massive impact on the quality of a match maker and MMR system,** you do not want to ever do what I just did and have it be random. 1 is a positive value because you won the game, this number again was randomly selected by me. 0 was used because I believe that's what the system would use if it feels a game was 100% fair... Now I might be wrong about that, but honestly 100% fair basically never happens so even if i goofed up its fine to just treat this as another arbitrary number im using here. Like i said not an expert so i will try to account for personal mistakes. There is are few ways to make We.. I don't know what riot uses so I wont get into it to much... because id just explain a lot that means very little. Apard on the topic of _We _provided 2 formulas for its creation. We= N X P(Dc) is one, We = ΣPi is another.
Why I am using made up numbers here to illustrate the adjustment being used because variables are hard for most people, myself very much included.** What is awesome is that the concept is actually pretty simple and easy for any one to understand. After playing a game if Win you go up in rating if you Lose you go down. **
As players, our rating is not meant to be some law or absolute measurement of skill but rather you can take it to be a Range of skill. When accurate its ment to set an expectation for your averaged performance of bad and good games. Humans are expected be inconsistent. Because people can improve (and in league people can get worse as well, game balance constantly shifts by paches) you never lock a player into a MMR. You let it always be able to move so that trends in playing result in shifting players the direction they trend.
K factor is a variable, its made to change over time. When its a very high value, you see people move around more dramatically. Like in my example, if k was 100 that player would be Rn = 1500. but a loss would be 1300. K changes largely for 2 reasons. Some ones up higher on the ladder into a MMR range you think players are becoming skilled, you lower it so that folks don't get to far up off luck. The other one is because people have played decent # of games. you now feel that they are nearing the MMR range that represents their skill and you don't want them to climb or sink as fast so you can consistently place them into games using MM around their skill.
As each win adds to your balance and each loss removes, what you see is events like this. A player with a 75% win rate! who played for all of 4 games? Ro + 36, Ro +36, Ro - 36, Ro + 36. Where you have the loss, this does not matter all to much because the ending mmr is going to be +72. One loss and one Win cancel out (not a realistic expectation, there is a good reason We is used in the adjustment. but this is fine in the boundaries of my example.)
Meanwhile some one with a 52% winrate but who played 500 games? Net gain of 2 wins per 100 played so that guys looking at 10 wins.
Comparing the two examples using my random starting value of 1400. 75% 4 game broski = 1476. The 52% 500 games played gent however is sitting at 1760.
Everything's way more complicated then this, don't treat my example as remotely relevant to what riot is using. _Its just to show how the system rewards consistency of play far more then a small sample size of games or what order a win/loss comes in.
_
So some basic theory on the topic of MMR. Your call over how reliable you feel I am as a source of information. In regards to Riot cant promise you accuracy but feel free to ask me any questions you'd like specifically addressed.
