### Why did Riot choose this ranking system?!

vWerewolf

Junior Member

Okay so yes I am a little bit angry right now because after 3 4v5 games in a row we just got matched with foreigners who insta lock a role we don't need and don't speak then halfway through the game start speaking english (Trolls). My friend and I always get forced into the worst roles and Still! do the best on our team.

My question actually is, Why on EARTH do Riot use a ranking system which is used in two player sports such as chess to measure relative skill. News flash there are 10 people!, when you are doing solo queue you still don't have communications with your team and if your below 1000 elo good luck even getting them to type let alone call mia's etc. My point is, after having such a terrible terrible run of 4v5's afks and trolls why do we continuously get punished for others leaving and for terrible runs. How does that measure My Friend And I's Relative Skill?

yagis

Member

P(any given 800 ELO player is a troll/AFK) = X

Then the probability that your team has no troll, assuming you are not a troll = (1-X)^4, and the probability that the other team has no troll is (1-X)^5. If you're duo-queued, that changes to (1-X)^3.

So if, say, 20% of low ELO players are trolls, but you are not, your team gets a troll 59% of the time.
Here's the issue: Your opponents get a troll 67% of the time.

Assuming equal P(Victory) in matches with equal number of trolls, and 100% chance of victory with a troll advantage(Which doesn't actually happen mind you), the following results, repeating that 20% trolls figure:

Case 1: You have a troll, your opponents do not. You lose 100% of these under the ridiculously inaccurate conditions earlier set. This comprises .59*.33, so 19% of games. Congratulations, you automatically lose 19% of your games. Sorry to disappoint you, you won't go undefeated.

Case 2: You have trolls, but your opponent does too. These split 50/50 if skill levels are even, and comprise 40% of the games.

Case 3: You don't have a troll, and your opponent doesn't either. 14% of games, split 50/50.

Case 4: Your opponent gets the troll, and you don't. 27% of games, and you win all of them.

So by the end of this, you win 54% of games just by being an average player and not a troll. This is obviously not the actual case, but that has a lot to do with the fact that you probably aren't as good as you think you are. Statistically speaking, 50% of the population has to be below 1200 ELO for it to be a true average.

ELO's a good ranking system; admittedly, Riot could do a better job of forgiving more games where there's an obvious AFK, but still, the problems your alleging solve themselves by immersing yourself in data.