You cant just making a matchmaking system that goes off how many wins and losing someone has
Its not. Its wins losses and context of who your playing.
It uses a probability distribution model. That sorts players by setting expectations. 50% odds for players at the same rating is the basic expectation, it does not matter how high or low your # is. If your vs some one with the same # it expects 50%.When its wrong, then it changes your rating. If your beating people consistently you move up until you stop winning a majority. If you lose consistently you sink until you stop losing a majority.
You cant blame teammates, if you called it a RNG coin flip WTF BADDIES DECIDED GAMES 100%... Then law of large numbers will come into play. Cuz your on your team 100% of games. The odds are stacked in your favor if your rating is lower then your skill and its only bad players who decided a game. Your not bad? Then your never an issue for team. So 10 players, 5 v 5. Team A has you. 4 remaining slots vs 5 remaining slots. 1 game losing player? Higher odds to be on the enemy team. Should look like 61% to 39% odds to win if the game is decided by baddies rather then personal impact assuming your never the baddie.
This results in a stable slanted WR for you personally, that means that your only ever going to climb in the long run because of "bad players". Obviously this means that if there is a game winning player, some one high skill who single handedly decided games then the odds are higher the enemy team will have that player... But... well if that's your argument your literally telling people there are higher skill players then you, and that high personal skill impacts match results. The very point of a skill rating system.
A mathematical equation provides the same results with the same variables. Its fair, its the same every time. If your data plugs into the system, and some other person has identical data, you end up with the exact same rating. You don't get locked in place, it will always adjust to new data.
This kind of system has existed for a long time. They are proven to work. They are well documented. They have been peer reviewed for ages.
Riots has its issues. Duoing is badly implemented. But despite that, as long as any one is willing to invest enough time to put in the games to remove RNG from a low sample size it will correctly get you into a Range.
Here is a rant on various MM systems and Rating systems
