How in-depth of a response do you want? I can grab some graphs and charts if you like seeing visuals. Wont be able to do that till tomorrow thou as I am heading out to work after I post this.
The issue I have is that the starting point at the bottom makes it very difficult to build a curve that can be used for quality MM and the end goal should be a skill biased ladder not a time based leaderboard. If you cant sink, if you don't have uncertainty. An account with 9000 losses into a mix of wins and losses that resulted in +20 game balance looks exactly the same as a new account that won 20 games back to back. They would get used by MM as equivalent skill, show up in the dame range of "skill" Have the same rank. Meanwhile in MMR that -9000 guy is at the very bottom of the ladder, and a new account with 20 Wins would be 20 games worth of MMR above the starting MMR.
So it impacts MM a great deal if you don't let folks sink and treat the lower 50% of players as equivalent in skill for MM usage. Like if one day MM said all players under silver 1 are now treated exactly as the same skill as Silver 1.
Now you do see Iccup sort skill, because skills a real thing. Its naturally occurring. The fundamental idea behind Elo, use naturally occurring averages to find a value to help represent skill expectations. His idea is hard to disagree with, In a game of skill by definition skill matters. Humans are inconsistent, good and bad performances, so use an average for their skill value.
So Higher skill vs lower skill results in better odds for the more skilled player to win. If this is consistently untrue then the assumption of who was more skilled at the game was wrong, correct their values in response because proof of skill is from results of a game. The more dramatic the gap in skill the more dramatic the impact on WR. Like beating the snot out of your friends in smashbrothers then getting your ass handed to you online. 100 wins vs some one your expected to beat 100% is not impressive.
Under League's MMR/ranking system, you can only be that sure if the there is a difference of 4 or 5 tiers which isn't very accurate.
Nah, we just see titles. They are a distraction. MMR has expectations built from the gap in values that's a consistent set. Using Arpad Elo's then 1c=200 points. At any place on the ladder a 200 point gap = 756% to 24% odds in the Elo system.
What Riots probability's are or the values they use to represent them are hidden, its why I say seeing the MMR number is pointless . If you don't have the context of what is used to build them and what the system thinks the odds between values are its pointless. Gaining 50 points in 1 system could be worth 100 in another. Some chess sites use a higher starting value for Elo to attract players, Some one whos 1400 on site A is prefers to use site B because they are 1600 and thus feel better.... Yet they are still the same skill at chess.
So its not just 5 divisions. Its off the points. equal points = 50% odds, usually with a few points of wiggle room. Then for every % break down you have a point range going from 49% to 51% all the way to 0.01% to 99.9% So 1300 has the same % expectations to win or lose inverted by the gap. 1100 loses to 1300 the same amount that 1300 is expected to lose to 1500. The entire thing measures you vs the average of the range your in for MMR. And as it gets smaller sample size as you move away from he center those averages represent less players.
