CSGOs placement games only count when you win so you need 10 wins, this is better as it places you based on how well you did only when you win. Whether you got carried or carried yourself. This way your placement rank is based on YOUR skill and rating rather than your teammates.
Thoughts?
I could imagine the Wood 5 players diaries.... "Day 97: Current record: 3W, 445L. Still trying to reach Diamond. Have not been placed yet. Maybe riot sees my potentional and is waiting for that one amazing game where I carry as Bard top. It will happen. I will get placed eventually. I have not lost hope. All is well friends."
Hey guys im like 1/12 and 60 cs at 30 minutes. I know yall are winning but mah elo so I'm feeding rest of game, kk bye bye :)
League's system is based on a sample size of 10 from a Binomial distribution. CSGO is a Negative Binomial with r = 10
Spoiler Alert: they both have the same fundamental parameter, p, the probability of winning a given game.
The estimate for p in Riot's case is simply the number of wins out of 10. The estimate for p in Valve's case is simply 10 divided by the total number of games played.
Both of these converge to a normal distribution as n or r become large rather quickly. The reason it's 10 is specifically because that's the point at which both of these distributions really start to approach normality quickly.
Requiring 15 placement matches would be better than replacing the current estimate distribution with a right-tailed skewed distribution at low values of r that will simply give you the exact same outcome as before. The only reason I would fathom to do this is if you have a fundamental need to place people even lower than what Riot currently does. I don't know how CSGO's ladder works; maybe they do like the conservative estimate for placements. However, I see no reason to make the system MORE conservative when games are already 40-60 minutes long on average.
because you are supposed to play for win, not for your personal performance
I'm pretty sure that's not how it works. Nearly entirely sure the "10 wins" thing just exists to ensure that players get accurately placed - it doesn't ignore your losses or anything. I'm pretty sure CSGO does reward extra ELO to the MVP of the winning team, but that's the only thing besides your win/loss ratio that effects elo. Your personal performance does not matter at all unless you are the MVP on the winning team.
I think it should be based on how you do in game.. such as KDR gold earned ex.. that would eliminate the problem with trolls and getting placed with players way out of your skill level. a lot of people get carried to a rank they aren't even supposed to be in, if ranked worked like this I think it would be 100% better, in my opinion.
