how is this fair matchmaking? ( in ranked )

Linda de Mol·Jul 26, 2019·3 votes·1,331 views

https://i.imgur.com/kdt4VwO.png

4 of the enemies have positive winrate, and in my team im the only one with a positive winrate?

how is that fair?

6 Comments

Sona PingJul 26, 20193 votes

Because the game doesn't match around win rate, that would be very bad. Because the matchmaker doesn't know what champions everyone's going to choose before they pick them.

Linda de MolJul 26, 20191 votes

[{quoted}](name=Sona Ping,realm=NA,application-id=3ErqAdtq,discussion-id=BpTGgbGe,comment-id=0000,timestamp=2019-07-26T20:37:17.614+0000)

Because the game doesn't match around win rate, that would be very bad. Because the matchmaker doesn't know what champions everyone's going to choose before they pick them.

The game ended up being absolutely horrible. A real pain. and 2 of them said they were drunk. At 24 min they both gave Fizz & Jax 20 kills.

I play one Ranked this day, and it ends up like this. Yikes.

and ofcourse not surrendering even though its far lost. because "drunk having fun lolzies"

Sona PingJul 26, 20192 votes

Ouch. Sorry about that. It's unfortunate that the only thing to do with those is to play more matches to dilute that game. Still, it sucks that you had to deal with it though.

Kai GuyJul 26, 20191 votes

Like Sona Said. Outch.

Thing is... its inherent to PvP that players are a major factor in match quality. No way around this.

Lots of folks don't get probability very well which is fine. If you want to know why Bad players or Toxic behavior feels so fucking always its pretty easy to show mathmaticaly.

Even 10% of player population is a massive MASSIVE # when you consider that its teams of 10.

Because 10% can be worded as 1 in 10. if 1 in 10 is expected to be a negative player and you have 10 people? Yea.... your gonna see it all the dam time.

Visual Aid! https://www.wizards.com/dnd/dice/dice.htm

Try it! Grab 9 Dice 10. Any 1 side on a d10 = 10%. 9 represents other players as your a non variable in the equation. Your not toxic, then 0% an issue. Your always toxic? 100% toxic player on your team every game. Your a random % of your games a negative? your what ever % that is.

So we don't roll 10 dice. just 9. Pick any number 1 to 10. That's your Negative player RNG 10% distribution. Roll for about 5-10 sets. Go back and count the # of games ya had where 1 dice rolled your "Troll" sided.

Example. I pick 1 for my number. Going to add X to represent you as a player. Going to divide teams for 5 to reflect league.

X,5,9,8,7 | 10,9,10,4,2,+0

X,2,9,6,9 | 8,4,2,8,4,+0

X,8,1,1,10 | 10,4,5,3,2,+0

X,1,3,6,10 | 8,3,7,1,2,+0

X,6,8,4,9 | 1,10,10,6,2,+0

you can see, 5 "players" total are negative values. Looking at combinations, 5/45 = 1/9 = 0.1111= 11.11%. Slightly higher then the expected 10% value I set but its a small sample size so its reasonable. The toxic 11% how ever impacted 3 matches 3/5 = 0.6 = 60% Over half my games. Lets run 1 more roll.

X,4,1,8,10,2 | 4,6,5,8,+0

Bummer. Now its 4/6. Now accounting for impact. I only really care about when my team gets a 1. So far at 6 games I have had 50% of my matches decided by baddies... but if I increase the # of games the distributing will increasingly favor my enemies to get the extra baddied… and that's 5/9 odds they have it so over all your favored to win 11.11% from a game loisingly bad under skilled player being in a match.

Not much you can really do besides make X as high of a result as you can and that's all about personal impact and consistency of play.

Kai GuyJul 26, 20191 votes

Probability distribution for assumed WR that changes when incorrect is the current system. Meaning that the context of who you play is a factor. Additional K is a variable used to track uncertainty and its ran on an individual to induvial biases so your Experience, the total # of games played, is a factor on your rate of change.

Win rate is heavily influenced by skill, but it alone does not represent skill with out context of who your playing. I can beat many of my friends in Smash brothers, some of them I could likely 2v1 and not lose a single life (which would skew stats like damage and kda). Because of the gap in skill come across as a far superior player and in some circumstances will have a 100% win rate due to the skill gap between me and a friend. However, I can not compete in any high skill tournaments as the gap between myself and the people who are truly skilled at the game, they will make me look as bad as I make some of my casual friends. This gap of skill is represented as a "Class interval" Where the gap of skill is significant enough that you don't treat the players on the same level, one player is simply outclassed and the results are predictably unfair because of it.

First lets discuss MMR. With MMR the range MM uses is built not off what players have, but off probability expectations. Because its sorts players in relation to the principle that same skill is fair you see this assumption made, equal points = 50% odds. Becuase of this assumption and the way the adjustment formual is ran, when the result of the matches which consistently do not reflect this expectation? That's when accounts climb or skin. win more then 50% in your range consistently, or lose more consistently and your account shifts up or down the ladder.

Next. Why do you not want to exclusively use WR for MM? Lets say it try's to make teams have an equal 50% win rates rather then RNG from who ever is in an set value. Your winning, your probability expectation in MMr means that you should climb and be move to a higher range of play for MM... your vs people who have numerically similar probability expectations from their results. So 60% WR in this context would be you have gained X points and moved up the ladder. Its now using your points to say, hey 50% fair if its in this Range, All players on each team are under a class interval gap so no ones grossly outclassed on paper from the game start.

But if you make it off WR? Yikes... To balance the team... the game looks for some one in that class interval but who has a gap that results in 50% or what ever is needed to balance the teams over all WR averages to be good. This is gross... your winning 80%? it has to cripple you. 3 50% and 1 20% for example. Or less dramatic but a handful of 30-40% to make the average lower for you.

so it punishes people who win. It helps people who lose. This is the exact fucking opposite of what you want on a ladder made from skill. You want bad to sink and good to climb. Forcing 50% is bad.

Plus your having to account for the fact there is a finite # of players at any given time. So your adding que times and widening the Range MM pulls from when there is not enough accounts in the Que.

So yea.. WR alone is not enough. Which is why Elo and MMR systems are more then just WR. Your adjustment is some deviation of Rn=Ro+K(W-We) The values are variables but the math remains consistent for each and every single player in its behaviors.

An example for simplifications. X+Y=7. You can have many different values of X or why to get the result yet the result is the same no matter what. MMR does this for a range of skill that represents a mean average, and it takes some time to get to low uncertainty which is when it shows its value.

Any questions?