Do you like these ideas?

Yes 20 46.51%
No 13 30.23%
I like Most 6 13.95%
I like part. 4 9.30%
Voters: 43. You may not vote on this poll

Thoughts on Elo/Ranked.

Comment below rating threshold, click here to show it.

adc

Member

12-08-2012

Quote:
Originally Posted by LogicalTautology View Post
I'm going to separate my post into individual steps so we can see where the difference of opinion is. Tell me the first statement that you disagree with.

1) Elo rating is designed to give a rough idea of player skill
2) Elo rating is not designed to be accurate over a small sample size
3) The goal of League of Legends is to win the game
4) If player A is a better player than player B, player A should have a higher Elo
5) If player A and player B both play some number of games at the same Elo level, and player A wins 51% of his games with an average KDA of 0/99/0 and player B wins 50.9% of his games with an average KDA of 99/0/99, then player A is a better player
6) Player A should have a higher Elo than Player B despite a much worse KDA, based solely on a higher winrate
7) Winrate should be the only thing that affects Elo, not KDA
I'd disagree with 5/6. Player A might (must?) have been carried by a significantly better player. With scores like those, I think most people would agree that Player B should have a higher win rate than Player A, given those KDs :P.

If it was more like 3/5/3 versus 5/0/5, though, that would make more sense.


Comment below rating threshold, click here to show it.

LogicalTautology

Senior Member

12-08-2012

Quote:
Originally Posted by AD Bottom View Post
I'd disagree with 5/6. Player A might (must?) have been carried by a significantly better player. With scores like those, I think most people would agree that Player B should have a higher win rate than Player A, given those KDs :P.

If it was more like 3/5/3 versus 5/0/5, though, that would make more sense.
The only constant in both player's games is that A is playing in all of his games and B is playing in all of his games. By definition it is impossible to get carried to a higher/lower Elo over a significant sample of games. Believe it or not, there is no doubt that the guy going 0/99/0 is adding more to his team's chances to win than the guy going 99/0/99 if he has even .5% more wins over a significant sample at the same Elo level.


Comment below rating threshold, click here to show it.

adc

Member

12-08-2012

That makes no sense. I can get a 2500 player to carry a 300 player to 1500, despite the 300 player going 0/15 every game. Simply because the 2500 will carry harder than a 1500 player will carry over a 300 player.

If you're talking solo queue, then I would probably go hang myself if a 0/99/0 average player got to a higher level than a 99/0/99. Your numbers are severly screwed. I get the idea behind it, but pick realistic numberse.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-08-2012

Quote:
Originally Posted by AD Bottom View Post
With scores like those, I think most people would agree that Player B should have a higher win rate than Player A, given those KDs
That's kind of the point though, isn't it? I mean, granted, 99/0/99 and 0/99/0 are EXTREME examples that will most likely never be attained, but if over a sample size of say 200 games, if Player A has a higher win-rate than Player B, then what else can we conclude other than the fact that Player A is better at winning games than Player B?

What is Elo supposed to be a measure of? When we look at point 3 and point 1, Elo is a measure of skill, and in LoL skill is translated as one's ability to win games.

If I have a 67% win-rate and you have a 52% win-rate, and we're playing against the same players, my ability to win games is greater than yours. What's more, if I have a 67% win rate and you have a 52% win rate, a team that includes four players with a 50% win rate plus me is more likely to win than a team that includes four players with a 50% win rate plus you.

And while it is true that getting kills and assists can help you win the game, this only goes so far, really. A player with 99 kills and assists is actually not any better off than a player with say 20 kills and assists. At a certain point, the player achieved a full item build and stopped getting stronger. In fact, during at least part of the game, getting more kills was arguably detrimental to his team because he already had a full build while his teammates were still working on their builds.

Meanwhile, even purely through passive gold gain, given enough time passing, you'll eventually obtain a full build. And let's suppose that 0/99/0 is a Tristana that happened to have perfect CS and her team killed dragon every time. She's building AD/AS and just suiciding on turrets. But the fact of the matter is, she's getting turrets killed. Heck, for all we know, none of her deaths even fed the enemy team any gold because she only died to turrets, minions, or monsters.


But ultimately, none of that even matters. The point is, if I have a higher winning percentage than you, I am better at winning games. There's no metric you can possibly come up with that will more accurately tell us who is better at winning games then Win-Loss ratio.




EDIT: And yes, the argument is strictly for solo-queue. Whether or not others are duo-queued into your lobby makes no difference, but this works if you're queuing by yourself.


Comment below rating threshold, click here to show it.

LogicalTautology

Senior Member

12-08-2012

Quote:
Originally Posted by AD Bottom View Post
That makes no sense. I can get a 2500 player to carry a 300 player to 1500, despite the 300 player going 0/15 every game. Simply because the 2500 will carry harder than a 1500 player will carry over a 300 player.

If you're talking solo queue, then I would probably go hang myself if a 0/99/0 average player got to a higher level than a 99/0/99. Your numbers are severly screwed. I get the idea behind it, but pick realistic numberse.
In theory, if matchmaking does its job, then (assuming both the 2500 and 300 Elo players are correctly ranked) they should still only get a winrate of approximately 50% by Duo Queuing. I think so, anyways. The idea should be that the 2500 Elo player only wins his lane by an equal amount to the amount that the 300 Elo player loses his lane by.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-08-2012

Quote:
Originally Posted by LogicalTautology View Post
In theory, if matchmaking does its job, then (assuming both the 2500 and 300 Elo players are correctly ranked) they should still only get a winrate of approximately 50% by Duo Queuing. I think so, anyways. The idea should be that the 2500 Elo player only wins his lane by an equal amount to the amount that the 300 Elo player loses his lane by.
In theory, this works.

In practice, I don't think it's entirely true. A 1400 will have no problem crushing a 300, and a 2500 will have no problem crushing a 1400. But now what do you have?

Suppose everyone but the 2500 and 300 are 1400, and the 2500 played as ADC, so it's harder for him to leave lane.

You have three 1400 on each team that broke even. You have a 2500 on team A that's fed, and a 1400 on team B that's fed. You have a 300 on team A that got zoned, and a 1400 on team B that got zoned.

Well, a zoned character is just about equally useless no matter what Elo you are.

But now, once we start team-fighting, we have a fed 2500 up against a fed 1400. The advantage is pretty clearly in favor of the 2500's team, I think.


But what's more, if the 2500 played mid or jungle instead, where his impact is maximized, while the 300 played support, where his negative impact could be minimized (and the ADC is baby-sitting him), then the 2500 is going to insure everyone on his team gets fed a bit by roaming and ganking. This is an even larger advantage for the 2500's team.


So anyway, for simplicity sake, we should keep this to talking about solo-queue for now.


Comment below rating threshold, click here to show it.

Alythezon

Junior Member

12-08-2012

My only problem with your "mitigating" is that it doesn't take into account how you help your team. Say I steal baron 3 times, but die each time. I helped my team a bunch, but it could affect my K/D/A therefore make it "unprofitable". That is an extreme example, but there are way more things other than K/D/A. You trade a 2 for 1 to get your adc fed when you are support, but it adds 1 to your death.

With such a large user base of LoL, I think the only real solution is that you need to find a team to play on and have enough players that you can always do 5's.

Also, Have you seen the pro teams play. I have seen a number of games where a team is way behind in kills >4, but they are ahead in gold. CS is more important than kills.


Comment below rating threshold, click here to show it.

adc

Member

12-08-2012

Actually, alythezon, I'd wager that most of the difference isn't CS, but objectives. One dragon is worth about two kills. Baron? Closer to 3 kills with assists. Towers are about the same as dragon (a bit less). CS is certainly part of it, but never underestimate objectives.

Now, the difference between kills and objectives is that kills concentrate your gold. I'd rather have a 6/0 ADC than get two dragons, assuming all else is equal (also, this may be the wrong choice, but it's my preference). Why? Well, a 6/0 ADC is an ADC with an extra 1800 gold, plus 50% of that in assists to whoever helps. Dragon is about 200 gold to each player on the team, or 400 gold each. So you're talking everyone gets 400 gold, or your ADC gets 1800 gold. It's situational, but most of the time I'd rather a fed ADC.