Elo hell? Elo is a horrible rating system? Someone help.

Comment below rating threshold, click here to show it.

TikalKeria

Senior Member

09-18-2012

I think it is a bad system >.<

Elo was created for chess which is a solo game :/

but LoL is a team game....it's a flawed rating system.

dunno what can be done though *shrugs*


Comment below rating threshold, click here to show it.

IncendiarySpade

This user has referred a friend to League of Legends, click for more information

Senior Member

09-18-2012

its the best system possible though. going by kda would screw certain roles, cs same thing, win percentage takes your ability to play as a team and win as well as your individual skill.


Comment below rating threshold, click here to show it.

MarlaMazzyHope

Member

09-18-2012

Quote:
Originally Posted by Sweetea17 View Post
I want you to ask yourself this question, if your stats were
568 wins in solo que
against 27 losses.

What kind of elo rating do you think you would have? 2700? 2200 easy right?

Not this guy,

http://www.lolking.net/summoner/na/24972085#profile
Just take a few minutes for that to seek in, get a soda, Smoke up.
Because, What is this?








This by far is the most disgusting thing I've ever seen.
For every champion this guy has played over 10 times he gets a 90 percent win rate.
Also his K/D is stupidly high


near 600 wins verses almost 30 losses and you get a nice near 1600 elo.


On the other hand. I show you a pro.
http://www.lolking.net/summoner/na/19807803

What is going on here. Someone please explain this to me, in complete detail.

Couldn't he be duoing with a lower elo friend, or pulling people up? That would minimize the positive elo gain for wins and stiffen the punishment for a loss, or am I wrong?


Comment below rating threshold, click here to show it.

Sweetea17

Junior Member

09-18-2012

Quote:
Originally Posted by IncendiarySpade View Post
Queue Dodging at beginning of season and/or duoing with someone who has 0 elo. Begone ye troll.
Lots of people did this 2+ months ago so they could have ridiculously high winrates. Some guy had like a 98% winrate with shaco at 45 elo. Anyone with half a brain knows what he did... so please don't shake the "elo hell exists" crowd's cage.
Quit refuting one side of this statement
This person is actively trying to find a Pro team
Read everything before you post please.
This isn't whatever you heard, Why is it so hard to pay attention?
This person is stuck.


Comment below rating threshold, click here to show it.

Sweetea17

Junior Member

09-18-2012

Quote:
Originally Posted by Tylerpb View Post
Dodging doesn't affect Elo anymore, but that's only been the case for a couple months or so while season 2 has been going on for almost a year now. You can't see a lot of his history, but it doesn't suggest that he is playing ranked games very actively. Queue dodging could still easily be the explanation.

It's really pretty hard to explain this based on some flaw in using Elo or applying it to a team situation or even Elo hell as the term is generally used. It's not like the guy is playing really well but keeps losing because of other people. He's winning a lot if that information is accurate. The only way your Elo stays down through that is by getting matched up with people way way below your Elo, which seems unlikely on that scale.

For what it's worth, the idea that doing well in a game should be correlated with getting a lot of Elo points for winning is flawed. You do really well in a game because you're much better than the people you're up against. If you get a small number of Elo points for winning that game it's because your Elo is much higher than those people you are much better than, as it should be. What have you really proven about your skill by beating someone the system already thinks you are much better than badly? What should get you a large number of points is when you beat people with a higher score than you, showing that your own score should be closer to or higher than theirs. The system gives a lot of points in order to move your score towards theirs, as it should be. In that situation it's a lot more likely that you'll have a very average score and barely eke out a win.

I honestly think they should add a type of ladder system along with elo, the ladder system would try to keep your win rate to 50 percent, so if its stupidly like with his jarvan 97 percent, It should get matched with MUCH better players.


Comment below rating threshold, click here to show it.

Sweetea17

Junior Member

09-18-2012

Quote:
Originally Posted by IncendiarySpade View Post
Queue Dodging at beginning of season and/or duoing with someone who has 0 elo. Begone ye troll.
Lots of people did this 2+ months ago so they could have ridiculously high winrates. Some guy had like a 98% winrate with shaco at 45 elo. Anyone with half a brain knows what he did... so please don't shake the "elo hell exists" crowd's cage.
Based on this data, Elo hell does exist, as an anomaly.
Please, Just don't be so blind.


Comment below rating threshold, click here to show it.

Sweetea17

Junior Member

09-18-2012

Quote:
Originally Posted by MarlaMazzyHope View Post
Couldn't he be duoing with a lower elo friend, or pulling people up? That would minimize the positive elo gain for wins and stiffen the punishment for a loss, or am I wrong?
It doesn't minimize for your WIN, It minimizes for the other teams win.

http://www.youtube.com/watch?v=1_xLjI6zN3I

I doubt that very much, If you are a 1500 player and your friend is a 0 elo player, It doesn't match up to 0 elo, For all of those people even thinking that is what it is.
I swear, So many people just don't do any research and just spit off whatever they think is right.

If I'm 1500 elo, and my friend is 500, That friend gets thrown into a 1500 bracket,
My source of information is, Phantoml0rd, when he does lessons and wants his student to play against high level play, he ques with them.
In other words, the math that goes into elo, Isn't OH YOU HAVE LOW LEVEL PLAYER YOU GET 1 elo, Its in fact the opposite


Because you are crippling your elo score as a team so you would actually end up getting more elo for each win, 680 wins, 27 losses.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

09-18-2012

Well... this idea that "Elo was designed for chess and doesn't work for LoL and here's an example!" is the worst possible imaginable argument when considering this evidence. The argument that Elo was designed for 1v1 games and not 5v5 games only works if you're trying to argue that the player in the team system is playing significantly better than the other 9 players in the game, but loses matches due to his teammates. IF that were the argument here, then the "Elo is for chess" argument would hold some water (not much, but some).

But that's not the argument here.

The argument here is that this guys has an exceptionally high win-rate, but is "stuck" in a surprisingly low Elo. In ANY game using an Elo system, you're awarded positive Elo for every win and negative Elo for every loss. Additionally, the amount of Elo gain/lost is adjusted based on the expected results of the match, as determined by the Elo of each team/player. So if Team A has significantly higher Elo, they'll get significantly less Elo for a win and lose significantly more Elo for a loss.

Now, there are only three ways to explain this guys Elo.

a) Queue dodging. No, it doesn't give negative Elo any more, but it used to, and it wasn't that long ago that it did. This guy could have easily been regularly queue dodging his Elo down to keep it low. The fact of the matter is, there's not enough data available on Lolking to prove this one way or the other, so the Lolking data isn't conclusive enough in this regard to say that LoL's Elo system is definitively broken--their simply IS NOT enough data.

b) Duo-queuing with someone with significantly lower Elo. If this player played every single match in a duo-queue with someone with extraordinarily low Elo, his team would likely have extraordinary expected odds of winning every single match. Therefore the system would be rewarding him extraordinarily low amounts of Elo for wins, while simultaneously penalizing him with massive Elo losses when his team didn't win. We could look at Lolking to check if any of the names in his last 10 matches are repeated, we could look at the Elo of the players on each team to see if there is a case where his team has abnormally higher Elo than the other team in the past 10 matches, however... this would only give us information about 10 matches out of 595 matches. So even if his 10 most recent matches are legit, that'd only account for like +140 Elo tops if they were all wins in cases where his team was actually expected to lose. There are 585 matches to look at that we have no data on, so again, we simply don't have enough data to conclusively, definitively say that LoL's Elo system is broken.

c) This player has the absolute worst luck and the random match-making system always puts his team at a massive advantage in terms of team Elo and expected outcome of the match, forcing a similar situation to the above situation, where wins are worth next to nothing and losses cost a lot of Elo, but this is essentially statistically impossible.


But regardless of anything, all that Lolking tells us is the player's overall win rate, and what his last 10 matches are. None of the last 10 matches were even ranked matches, so we actually have zero specific data on any of the ranked matches this player has played. There is no conclusive evidence that says anything about what is making this player's Elo what it is, one way or the other. There are a lot of explanations for him having this Elo and this win-rate, and the only argument the OP has against these explanations is that this player was actively seeking to join a pro team, and we're supposed to take his word as definitive proof that the Elo system in LoL is broken. No thank you.


Comment below rating threshold, click here to show it.

Sweetea17

Junior Member

09-18-2012

Quote:
Originally Posted by PogoPogoPogoPogo View Post
Well... this idea that "Elo was designed for chess and doesn't work for LoL and here's an example!" is the worst possible imaginable argument when considering this evidence. The argument that Elo was designed for 1v1 games and not 5v5 games only works if you're trying to argue that the player in the team system is playing significantly better than the other 9 players in the game, but loses matches due to his teammates. IF that were the argument here, then the "Elo is for chess" argument would hold some water (not much, but some).

But that's not the argument here.

The argument here is that this guys has an exceptionally high win-rate, but is "stuck" in a surprisingly low Elo. In ANY game using an Elo system, you're awarded positive Elo for every win and negative Elo for every loss. Additionally, the amount of Elo gain/lost is adjusted based on the expected results of the match, as determined by the Elo of each team/player. So if Team A has significantly higher Elo, they'll get significantly less Elo for a win and lose significantly more Elo for a loss.

Now, there are only three ways to explain this guys Elo.

a) Queue dodging. No, it doesn't give negative Elo any more, but it used to, and it wasn't that long ago that it did. This guy could have easily been regularly queue dodging his Elo down to keep it low. The fact of the matter is, there's not enough data available on Lolking to prove this one way or the other, so the Lolking data isn't conclusive enough in this regard to say that LoL's Elo system is definitively broken--their simply IS NOT enough data.

b) Duo-queuing with someone with significantly lower Elo. If this player played every single match in a duo-queue with someone with extraordinarily low Elo, his team would likely have extraordinary expected odds of winning every single match. Therefore the system would be rewarding him extraordinarily low amounts of Elo for wins, while simultaneously penalizing him with massive Elo losses when his team didn't win. We could look at Lolking to check if any of the names in his last 10 matches are repeated, we could look at the Elo of the players on each team to see if there is a case where his team has abnormally higher Elo than the other team in the past 10 matches, however... this would only give us information about 10 matches out of 595 matches. So even if his 10 most recent matches are legit, that'd only account for like +140 Elo tops if they were all wins in cases where his team was actually expected to lose. There are 585 matches to look at that we have no data on, so again, we simply don't have enough data to conclusively, definitively say that LoL's Elo system is broken.

c) This player has the absolute worst luck and the random match-making system always puts his team at a massive advantage in terms of team Elo and expected outcome of the match, forcing a similar situation to the above situation, where wins are worth next to nothing and losses cost a lot of Elo, but this is essentially statistically impossible.


But regardless of anything, all that Lolking tells us is the player's overall win rate, and what his last 10 matches are. None of the last 10 matches were even ranked matches, so we actually have zero specific data on any of the ranked matches this player has played. There is no conclusive evidence that says anything about what is making this player's Elo what it is, one way or the other. There are a lot of explanations for him having this Elo and this win-rate, and the only argument the OP has against these explanations is that this player was actively seeking to join a pro team, and we're supposed to take his word as definitive proof that the Elo system in LoL is broken. No thank you.
Not definitive proof and not just my word, Anyone who was watching the Stream with TSM Theoddone saw this guy and listened to his whole story.

I plan on tracking his person for a month or so to gather more data, also, Queing with someone brings them "UP" not down, unless it depends on the host, If it does, I wasn't aware of that.
I am not arguing anything, I'm stating these, as far as I know are facts, All I'm trying to do is find an answer, My mind isn't "THIS ELO SYSTEM SO BROKEN OMG OMG ROIT YOU SUCK BLAH BLAH"
Its actually more, Hmmm this doesn't seem correct, What the hell? Must find more answers.

aside from that, Thank you for the in depth explanation, its been by far the most informational of the responses.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

09-18-2012

Quote:
Originally Posted by Sweetea17 View Post
Not definitive proof and not just my word, Anyone who was watching the Stream with TSM Theoddone saw this guy and listened to his whole story.

I plan on tracking his person for a month or so to gather more data, also, Queing with someone brings them "UP" not down, unless it depends on the host, If it does, I wasn't aware of that.

aside from that, Thank you for the in depth explanation, its been by far the most informational of the responses.
There are still other explanations.


And as I said, there's no ranked matches in his 10-match history. In fact, he hasn't played a ranked match in the last 7 days even. Moreover, his Elo has remained a constant 1493 since 09/05/12, so unless he's gained and earned the exact same amount of Elo in the same day, he probably hasn't even played ranked in the last 13 days even.

And while this still isn't definitive or conclusive, it suggests that he definitely hasn't been playing much recently. And that makes it all the more likely that he was able to inflate his win percentage by queue dodging back when that provided an Elo penalty. Again, there's no direct evidence, but to me, that's what it points too. Get him to go play a ranked match today. If he wins, watch him get 10-14 Elo.