Why we hate Elo-Hell believers

Comment below rating threshold, click here to show it.

deeeJAY

Member

01-03-2013

I understand how Elo cannot be manipulated so that players recieve better placement in the rankings, for fear of abuse. I also know that again, regardless of how it is currently working or if its semi-efficient at getting the job done, its a broken system for a team based game. If i place a bucket under a leaking roof, yeah the bucket will collect water...but if the bucket has a hole in it, is there really a point to putting a bucket there, or should i replace the bucket with a bin, or one without a hole?

Quote:
Originally Posted by Selcopa View Post
Under your understand of elo, a team 200 points higher will ALWAYS beat the lower team, 100% of the time, because 1 team is higher therefore they wont lose.
If the system was accurate, then yes this should be not only probable but factual. At least for 5v5 ranked. For solo ques and duos? Maybe not.

Edit: 200 points may be a bit low for 100% but I'm sure you get the general idea.


Comment below rating threshold, click here to show it.

Selcopa

Senior Member

01-04-2013

Quote:
Originally Posted by deeeJAY View Post
I understand how Elo cannot be manipulated so that players recieve better placement in the rankings, for fear of abuse. I also know that again, regardless of how it is currently working or if its semi-efficient at getting the job done, its a broken system for a team based game. removed stupid comparison
You can call it semi-efficient if you want, we'll agree to disagree on that one, you have no evidence of it being inefficient that you've actually brought forward other than the QQ's of people who are bad. however one thing is certain, there is not a single system that is more efficient at placing players relative skill than the elo system, team or individual. Is it more effective in 1v1 competitions? Absolutely but no system is more effective in 5v5 than in 1v1, even systems designed for 5v5



Quote:
If the system was accurate, then yes this should be not only probable but factual. At least for 5v5 ranked. For solo ques and duos? Maybe not.

Edit: 200 points may be a bit low for 100% but I'm sure you get the general idea.
I see the misunderstanding, this aspect of your post is direct evidence that you do not understand an elo system, allow me to quote the concept so you can modify your opinion.

"It is assumed that a person's performance varies from game to game in approximately a normal distribution and a person's Elo rating is the mean of that distribution. A person with a higher Elo may perform better on average than a player with a lower Elo"

"For every difference of 400 points, the team/player with the higher score is ten times as likely to win as the other team/player. This standard is for Chess and may be different in League of Legends"

this means, that even in situations where the team average 200+ points higher than their opponents, when the system is perfectly accurate there still is a chance for the lower team to win.

This is especially true in team based games using the elo system as multiple players creates significantly more variables than a 1v1 environment


Comment below rating threshold, click here to show it.

Larrythetapist

Senior Member

01-04-2013

Quote:
Originally Posted by deeeJAY View Post
http://www.census.gov/newsroom/relea...cb11-ff15.html
According to that site, 19.7 million students attended college in the 11-12 school year.
http://theivycoach.com/2012-ivy-leag...ns-statistics/
According to this site, a mere 23,514 students were accepted into ivy league schools in 2012.
That is .11% of the entire collage populace.
http://leagueoflegends.wikia.com/wiki/Elo_rating_system
says Gold rankings are the top 5-1%. Since .11 is lower than 1 this would INSTANTANEOUSLY make them a platinum or higher player.
So you admit most of the people in the ivy league are the cream of the crop i.e. the real life Elo Hell doesn't exist and people who attend good schools have high-elo equivalent intelligence.


Comment below rating threshold, click here to show it.

deeeJAY

Member

01-04-2013

Quote:
Originally Posted by Larrythetapist View Post
So you admit most of the people in the ivy league are the cream of the crop i.e. the real life Elo Hell doesn't exist and people who attend good schools have high-elo equivalent intelligence.
George W Bush was DEFINITELY the cream of the crop. You got me there!

Edit: Weren't we JUST talking about legacy admissions? Please stop posting.


Comment below rating threshold, click here to show it.

deeeJAY

Member

01-04-2013

Quote:
Originally Posted by Selcopa View Post
You can call it semi-efficient if you want, we'll agree to disagree on that one, you have no evidence of it being inefficient that you've actually brought forward other than the QQ's of people who are bad.
"Why i hate Elo-hell non-believers: http://i.imgur.com/ClTrC.jpg"

I posted this picture awhile back. I noticed you failed to comment on it. Unfortunately Lolking is down so I'm unable to re-check how he's currently doing. This same thing happened to me and many others. How can you call this fair, balanced or even remotely accurate?

Quote:
Originally Posted by Selcopa View Post
This is especially true in team based games using the elo system as multiple players creates significantly more variables than a 1v1 environment
lol? I'm at a loss for words here. You're arguing my point here and you're unaware. Yeah the system works flawlessly for 1v1, but as you've mentioned here, there are a variety of other variables that aren't taken into account. So how can you say elo is a correct presentation of skill and not some amount of luck? (refer to the image above) Especially for solo and duo queued rankings, where there are 3-4 other people you yourself cannot account for, yet they are a representation of YOUR skill level. At least, according to Elo.

Edit: Forgot to mention, you brought up abusing the system awhile back. Is it not possible to abuse the system into getting placed higher, IE selling Elo? There's always going to be a way to jeopardize the integrity of any system. The question is: are you willing to work around it?

For Riot the answer is most likely "no". No one has to time or patience (potentially the knowledge) to come up with a mathematical formula to fix this problem. As long as they continue getting paid and the system somewhat works, who cares right?


Comment below rating threshold, click here to show it.

corallein

Senior Member

01-04-2013

Quote:
Originally Posted by deeeJAY View Post
lol? I'm at a loss for words here. You're arguing my point here and you're unaware. Yeah the system works flawlessly for 1v1, but as you've mentioned here, there are a variety of other variables that aren't taken into account. So how can you say elo is a correct presentation of skill and not some amount of luck? (refer to the image above) Especially for solo and duo queued rankings, where there are 3-4 other people you yourself cannot account for, yet they are a representation of YOUR skill level. At least, according to Elo.
Your Solo Queue Elo is representative for the level of your play for Solo Queue games. It is largely irrelevant who the other players in the game are, as it measures your ability to win games in Solo Queue. All other statistics are unimportant in this context - kills, deaths, assists, gold earned - none of them are directly relevant to your win rate and Elo.

The usage of skills is different in all game-modes, from game mode (Dominion, Twisted Treeline, Summoner's Rift), pick mode (Blind vs. Draft), and Solo vs. Team Queue. For instance, a player who can only play one role well is likely going to do better in Team Queue in a team where he/she gets to exclusively play that role than in Solo Queue.

Quote:
Edit: Forgot to mention, you brought up abusing the system awhile back. Is it not possible to abuse the system into getting placed higher, IE selling Elo? There's always going to be a way to jeopardize the integrity of any system. The question is: are you willing to work around it?
"Selling Elo" is fundamentally different than gaming the system, since it involves one player masquerading as another. This is directly cheating the system, which only works when Elo is only associated with one player. Not to mention that it's unsustainable and the player who "bought" the Elo will quickly lose it once they start playing their own account again.

There are other ways to game the system, such as duo queuing and smurfing. But again, those are different in that they're both unsustainable, and both also fundamentally break the system. Yet the only way to prevent those would be to prevent playing with friends (no duo queuing in Solo Queue) and only allow a single account per user (eg. tying account creation to a real-world identifier of some sort) respectively. Riot's not going to do either, as both of those actions would be tremendously unpopular.

Quote:
For Riot the answer is most likely "no". No one has to time or patience (potentially the knowledge) to come up with a mathematical formula to fix this problem. As long as they continue getting paid and the system somewhat works, who cares right?
If someone could come up with a better way of measuring Solo Queue skill than Elo, I'm sure Riot (and other companies too) would love it. No one has yet.

The only reasonable proposal I've ever seen in these forums (in three years) has been using Glicko instead of Elo. But Glicko is just a modified version of Elo that should result in faster convergence to a user's "true" skill, but does nothing to solve the underlying issues of applying a skill rating system to a game where you have random teammates.


Comment below rating threshold, click here to show it.

corallein

Senior Member

01-04-2013

Also, that futurefreak person isn't very good. If you get that fed as Akali, you should absolutely have built a Guardian Angel. His 11 deaths (tied for most on his team) is an indication that the other team was doing a good job of focusing him. Or that he was diving in too early in fights and dying. Or he was getting caught out-of-position a lot and dying. And he has no Void Staff while the other team stacked a ton of MR.

Though nothing can be concretely determined without watching a replay... yeah, he doesn't seem that great.

Edit: Also, why the heck does no one but Darius have a boot enchantment in a 47 minute game.


Comment below rating threshold, click here to show it.

ninjagear

Senior Member

01-04-2013

why is it so hard to understand?

the more u win the more elo u get

the more u lose the more elo u lose

how is this broken? deejay just repeats what someone else says and then goes 'no ur wrong'

arguing with highschool students is so tiresome


Comment below rating threshold, click here to show it.

Ricardo MX

Junior Member

01-05-2013

Can't believe I read through all this ****.


Quote:
Originally Posted by deeeJAY View Post
George W Bush was DEFINITELY the cream of the crop. You got me there!

Edit: Weren't we JUST talking about legacy admissions? Please stop posting.

This guy is the only who knows what he's talking about. There are always going to be exceptions. Some smart people get gimped out of good schools because they didn't do good on the day of the SAT (placement matches).

And some retarded people get carried into the Ivy League by virtue of their family (duo q and previous-season seeding).

What do you guys not get? estupid americanos


Comment below rating threshold, click here to show it.

Selcopa

Senior Member

01-05-2013

Quote:
Originally Posted by Ricardo MX View Post
Can't believe I read through all this ****.





This guy is the only who knows what he's talking about. There are always going to be exceptions. Some smart people get gimped out of good schools because they didn't do good on the day of the SAT (placement matches).

And some retarded people get carried into the Ivy League by virtue of their family (duo q and previous-season seeding).

What do you guys not get? estupid americanos
While its stupid to turn this into a political thing, I will say that anyone who believes a president of the United States isn't a brilliant individual is politically biased and an idiot, this goes for President Bush(while he may not be a great speaker, the guy really is brilliant, read his book) as well as Obama.

But all of that is irrelevant, your argument is entirely false, you are trying to say that high elo players last season are high elo this season because they got seeded higher, this is just simply not true.

Over a short period of time? Sure maybe this season they get lucky, but just like a player who gets placed too low, over time if they aren't good enough to sustain the elo, they will gravitate and ultimately end up where they belong.

Think of it like this:
Higher Influence on Game -------------- Lower Influence on Game --------------Your TRUE Elo -------------- Lower Influence on Game -------------- Higher Influence on Game

As you move away from your true elo, your influence on the game goes up. When you are below your true elo, the farther away you are, the more likely you are able to dominate the game, as you approach your true elo, your ability to influence the game goes down because you cant take advantage of mistakes as well(or you make mistakes as well)

When you reach your true elo you have the least amount of influence on the game, this is because you aren't any better than your opponents, so its dependent on if you get bad opponents or good teammates. If you get some good luck with your matchups and move beyond your true elo, your influence in the game will actually increase, it will go up but in a bad way. Because you are above true your elo, you are more likely to make mistakes that are taken advantage of more frequently than your teammates will.

And the higher above your true elo you go, the more influence you will have because the more often you will make mistakes. You can still win with good teammates or in situations that you are stronger in, but you have a huge disadvantage