Every time this is mentioned here is what it boils down to. If you add stats the MMR and LP ratings STILL have to be attached to wins and losses.
How do you have Stats and not reward selfish players?
The system is not going to do a great job of subjectively grading players.
Stats not being subjective will lack the context of impact on a game, this blatantly sucks as an objective measurement of skill.
who casts his R every single time it comes off CD is going to have a much higher Damage from that ability then the one who saves it for the correct times. If that player is only rewarded for stats then he is incentivized to pad stats. Stat padding is very very easy. Spam ability's any time you have a target. Abandon teammates in fights if it looks like it hurts your KDA. It takes a lot less work to play for stats then it does to play to win. Your ADC is behind, needs a wave or two for a big item powerspike? The correct play is to let them have your wave. The Stat padder take every single minion they can, not because they want to win but because they are looking to have better stats.
So then your left with stats alone failing to help, so it goes back to wins and losses. Your "Analyzed performance data" is meaningless outside of the context of the match. You'd need a Subjective system that's looking at the match in the context of the match for it to function. We don't have the ability to automate that.
If I beat a player in Chess in 2 moves. This does not tell you how good I am. Just that my opponent was terrible and made the worst possible moves as white. You cant gauge my skill in that game. My stats are the very best possible 2 moves but the contact, the subjective rating is simple, that my opponent was completely ignorant of the game or wanted to lose.
The issue with MMR is most players don't understand it and wont learn. They think a win streak compensates them to climb, oblivious that the system functions more like a Checking book balance and not like a leaderboard. They see the top of the ladder, and think oh leaderboard righto. The leading players just have a higher balance however, its not something you can achieve just by putting in time/acquiring a high # of wins alone.
Your balance is made from some mutation of Arpad Elo's formula. Rn=Ro+K(W-We) <- this adjustment is the core for what your balance is.
But folks stop there and don't learn any more.
For example, most people are going to be completely lost when I say that to an extent It should be clear to people who studied Elo systems that in converting from 1v1 to 5v5 We variable expectations needs to be adapted. In chess We is made 1 of two ways. We +ΣPi Or the slightly less accurate but still workable We=Nx P(Dc) , Dc = (R-Rc) This is going to go over most players heads yet they sit here talking like they know the system or their opinion, built exclusively from their match history is going to be superior?
Anyhow Point being Stats are a bad way to fix Matchmaker. If you don't quite get MMR Matchmaker systems here is a visualization.
https://imgur.com/a/4SFSAf9
Forgive the MS paint skills im not an artist.
To my knowledge, This should be fairly accurate to Elo/MMR systems as a whole. I posted it a while back and have yet to meet any one who correct me so I am just going to roll with it as accurate until I learn new information stating otherwise, be that on my own or some one else providing it.
The problem is players assume that the assigned Ranges are instantly accurate either from game 1 or from a very short # of matches when there is an adjustment period to move accounts into the right ranges. Its longer then placements. MM quality is highest when all players have decent sample sizes. Soft resets assist in reducing the # of required games.
