ELO feed back

Comment below rating threshold, click here to show it.

lFalse Kingl

Member

12-09-2012

I am concerned about the ELO system. It is very inaccurate since it is based on teamwork. Because ELO is based on teamwork, it shouldn't be applied in match making nor it should match players based on winning percent. ELO should be based on individual skills and performance instead. This will separate the bad players from the average or above average.

Think of it as if this was an FPS game. If ELO was based on your winning chances, instead of K/D/A, the below average player would be matched with above average player. That means, players with .5 kda would be playing against players with 1 or 2 kd and above. This would create a complete unbalance over the game, just as it is doing to this one. Which is why I am suggesting ELO should be based on individual skills. That way, the good players get matched with good players and bad players get matched with bad ones. This will avoid the illogical system that can't determine whether players are good or bad but only match them randomly based on your winning percentage.

What methods could be applied that would be much better than this system? In my opinion, consider the following.

Things to note:
// Values can be changed more accordingly, this is just an estimate.

Instead of applying ELO points by winning, we do it by their overall performance in battle.

1.1 CS is dismissed, because support class can't get CS.

Assist, turrets/tower, inhibitors, and kills can apply ELO, death cannot.

The average overall score you gain per match will determine ELO, along with any other objectives that aren't listed.

ELO could also be applied to each roles instead. For example, adc elo will be based on kills and cs while support elo is based on healing/buffing adc, etc.

Any other options which are not calculated randomly work as well. I simply want the ELO system to be able to tell the difference between a player who performs good in game and one who doesn't.


Comment below rating threshold, click here to show it.

Flipsy

Senior Member

12-09-2012

I agree that the ELO system is crappy, but your suggestions are overall terrible.

Having assists, kills, turrets/towers, and inhibitors should never apply ELO because then it makes the game toxic (see Darius thread on how his ultimate makes the game toxic). ELO can't be applied to roles because you can't state what role you'll play, so Karma mid will warrant what, support or AP mid? You can't declare it on the ELO system. The ELO system would actually improve at showing a person's individual skill level if duo queue were removed. ELO by definition is a show of INDIVIDUAL skill level, so having duo queue defeats the purpose altogether. That's it. LoLN will be able to best dictate a person's true ELO after that, without you or someone else having a 2nd person to carry or be carried by.

My personal suggestion to Riot a couple months ago about this was to remove duo queue and allow a 1-5 rating of your teammates and/or enemies. You rate your teammates from best and most helpful to worst based on their gameplay, since only their teammates could see how well they played. Sure, this will initiate certain selfish behaviors and maybe a few people will vote for themselves every game, but I've had a lot of games as Orianna where my team admitted that if it wasn't for me, we would have lost. I got almost no kills, no deaths, but assisted heavily in all but 3 of our team's 31 kills. I'm fairly certain my team would have voted me as "MVP" of that game, and then go down the line from there.

What would the voting system do? Winning team's MVP gets the most ELO of the batch (but the spread wouldn't be huge, it'd be, like, 17, 16, 15, 14, 13). If there are ties, such as 2-1-1-1-0 votes, the spread would be like 17, 16, 16, 16, 15 so it's not THAT bad. This ALSO would help with trolls in Ranked games in my opinion. This system for the losing team would allow MVP's to lose less ELO, and have the person with the most downvotes gets down into "ELO hell" faster. I know we all have had that one game where a terrible feeder was reported (and would be reported anyway). You report the one person who you feel fed or assisted the enemy team, and the ELO spread would be losing 17, 15, 13, 11, 10 or something like that.


Comment below rating threshold, click here to show it.

lFalse Kingl

Member

12-09-2012

"Having assists, kills, turrets/towers, and inhibitors should never apply ELO because then it makes the game toxic (see Darius thread on how his ultimate makes the game toxic)."

Both kills and assist have equal value points. This is to stop ksing and promote teamwork. Turrets should be worth the double points since there aren't many and it's harder to take them out. This will encourage teams to both play as a team and aim for turrets while dis promoting kills/ks or negative play.

As for what you said about classes, each person in lobby gets to pick a role then select characters available to that role. Characters can be available to more than 1 role. However, I prefer the above suggestion before applying this one.

Wrong thread :S Miss click !


Comment below rating threshold, click here to show it.

Tha Dynamite

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

Senior Member

12-09-2012

First of all, wrong forum.

Second of all, you clearly didn't think this through.
Lets say I went 5-0-10 in a game. Someone else went 5-30-10. In your system, they get the same score.

How are points from turrets and inhibitors distributed? From the person that gets the last hit? That hardly seems fair.
Or do the whole team get it? If the whole team gets it, whats the point of even having it in the equation?

Average score per game? What if I play a low kill game? What if it was a surrender 20 game? What if I get 10 games in a row that was low kill and surrender 20s? Wouldn't my ELO drop?

So essentially, players that go in stupidly and try to get kills and die trying will have higher ELO than those that play smart and conservatively.

What about, I don't know, winning or losing? If ELO is based on average score, whats the point of winning? Why not try to go for 1 hour games with every single possible objective down? Why not have games where both teams agree to farm the entire game so they can pad their ELO?

Third of all, score in a game is all relative.
Going 20-0 in a low level game is not the same as going 20-0 in a tournament level game. In your system, it will be.
Whats the point of even having a system then.

Fourth of all, how is matching those with similar win rates at a certain level illogical and random? That is exactly how an ELO system works.
Based on your level of contribution to a game, you will win or lose. If you contribute more positively over time, you will win more over time. If you contribute more negatively over time, you will lose more over time.
If the system doesn't work, how come 2k players are so much better than 1k players?

Score in a game means very little. If you went 30-0 and you just stayed bottom and farmed while they pushed down your Nexus, you'll still have a good score but contribute absolutely nothing to the win.

Maybe you should play the game a little more and actually figure out how it works instead of honor farming for ribbons.


Comment below rating threshold, click here to show it.

Absolute 42

Senior Member

12-09-2012

The thing is, your skill as a player is not defined by your mechanics, or your K/D/A, or even how many wins you have.

It's a whole bunch of things combined, a lot of intangibles that lead you to be more likely to win than another player (that especially includes teamwork because that's what this game is all about).

That said, if you are a better player, you will rise in ELO because it measures everything relevant to the game in that it records whether you win or lose.

If you go 5/0/30, it means nothing if you can't carry your team to victory. I would much rather be a 0/2/30 support that made key plays and saved the team with my wards.

There are a lot of paths to victory. Since that's the only thing the ELO system measures (victory), then it gets at all of those intangibles like leadership, communication, decision-making, etc. which your system misses.

Your system would encourage selfish play, not team-oriented play.


Comment below rating threshold, click here to show it.

Flipsy

Senior Member

12-09-2012

Quote:
Originally Posted by lFalse Kingl View Post
"Having assists, kills, turrets/towers, and inhibitors should never apply ELO because then it makes the game toxic (see Darius thread on how his ultimate makes the game toxic)."

Both kills and assist have equal value points. This is to stop ksing and promote teamwork. Turrets should be worth the double points since there aren't many and it's harder to take them out. This will encourage teams to both play as a team and aim for turrets while dis promoting kills/ks or negative play.

As for what you said about classes, each person in lobby gets to pick a role then select characters available to that role. Characters can be available to more than 1 role. However, I prefer the above suggestion before applying this one.

Wrong thread :S Miss click !
Wrong, this is promoting a fragfest. This promotes negligent gameplay and severe cases of tunnel vision that will result in losing the game regardless. You can't gauge how a turret's worth in ELO could factor in, so that's wasted effort. I've seen AD LB and Crittlesticks in mid win games, sooooooooo again your choosing of roles is null and void at this point, that does nothing.


Comment below rating threshold, click here to show it.

lFalse Kingl

Member

12-10-2012

created new one in the right section http://na.leagueoflegends.com/board/...3#post32275943
reply there for feed back on my half sorted suggestion.


Quote:
Originally Posted by RamboRal View Post
First of all, wrong forum.
It's a the bottom, I miss click nor know how to move the thread.

Quote:
Second of all, you clearly didn't think this through.
Lets say I went 5-0-10 in a game. Someone else went 5-30-10. In your system, they get the same score.
Yeah, however, can they keep getting 5/10 each time? I do not want to apply death as negative elo to avoid players from taking risks, such as tankers. My elo is based on average score. The more positive your results are, the higher your elo, dismissing deaths. However, maybe 20 deaths could be apply as negative. This is not fully develop, just a proposal

Quote:
How are points from turrets and inhibitors distributed? From the person that gets the last hit? That hardly seems fair.
Or do the whole team get it? If the whole team gets it, whats the point of even having it in the equation?
The goal is to promote players to aim for objective. Whole players should get point and last hit should get most I would suggest.

Quote:
Average score per game? What if I play a low kill game? What if it was a surrender 20 game? What if I get 10 games in a row that was low kill and surrender 20s? Wouldn't my ELO drop?
Yeah, this is based on how much work you can produce. The main goal is to reward players who work hard instead of rewarding everyone. This will enable to the system to distinguish bad players from good.

Quote:
So essentially, players that go in stupidly and try to get kills and die trying will have higher ELO than those that play smart and conservatively.
You would have to be more specific. They don't get rewarded unless they manage to get a kill or assist.

Quote:
What about, I don't know, winning or losing? If ELO is based on average score, whats the point of winning? Why not try to go for 1 hour games with every single possible objective down? Why not have games where both teams agree to farm the entire game so they can pad their ELO?
This can stay, I am saying that it shouldn't be based only on winning/losing. Players who work hard should receive some benefits.

Jobs used to work how this system does. A bunch of lazy people can earn the same amount of cash as someone who work hard. Because they used to get paid about the overall work and not those who went beyond. I am applying the current work field method to this elo system.

Quote:
Third of all, score in a game is all relative.
Going 20-0 in a low level game is not the same as going 20-0 in a tournament level game. In your system, it will be.
Whats the point of even having a system then.
The one who goes 20-0 in the tournament will have a higher elo if they play against the lower level player.

Quote:
Fourth of all, how is matching those with similar win rates at a certain level illogical and random? That is exactly how an ELO system works.
Based on your level of contribution to a game, you will win or lose. If you contribute more positively over time, you will win more over time. If you contribute more negatively over time, you will lose more over time.
If the system doesn't work, how come 2k players are so much better than 1k players?
I do not know their methods and some ways of by passing 1.5k elo such as duo, winning games early(10wins-0 loss) or dodging(most common). Why? Because it doesn't matter if you get match with higher elo players. There's a chance those players are bad because elo does not determine their individual skill but their winning rate that could be based on other players effort. Duo also decreases the chances of being with feeders, afk, etc. Those who played early, as soon as the season reset the elo, have a higher chance of getting out of elo "hell"/randomness than those who play in a month or so.

Contributing a lot to increase your chances of winning does not change the fact that you are being unfairly matched. No matter the effort you put, you will be put with players who might not be able to play. For example, in fps games, there are matches which allows a certain amount of kd/elo. This separates the good players from the bad, unlike this one which works randomly since it is based on the overall winning percentage. I am sure you should understand, just consider how you can't distinguish a good player around elo 1300 or under I'd say.

As for why players with 2k might be batter than 1k? Not all 2k players play better than 1k. However, the majority will play better than the 1k. Why? You need basic statistics for this. How likely are players to remain under 1k elo? How like are good player to remain under 1k elo because their team drags them down? It is all based on probability. The truth is, there will be a lot of players doing bad and dragging down average or above average player. So you can expect 1k players to play bad since there will be many but there will always be good players who can perform as good as those with 2k elo or higher. What's the difference between them? Luck, one ends up with maybe average team while the other one doesn't. Even if one might be better than the other, the result will be determine by team play. Which is currently inaccurate since you get match randomly.

Quote:
Score in a game means very little. If you went 30-0 and you just stayed bottom and farmed while they pushed down your Nexus, you'll still have a good score but contribute absolutely nothing to the win.
30 kills is a high contribution. Especially since the chances of other players getting assist increases, etc. Even if the player just farm, winning also counts toward elo, which means it would be more accurate than players performance not counting.

Quote:
Maybe you should play the game a little more and actually figure out how it works instead of honor farming for ribbons.
Even if I played little, it takes no experience to figure this out. Based on the sources I gathered, elo is random. Doesn't matter how good you play per average, your chances will be determine by teams. At least with what I am suggesting, you will be put between players do can play at average or so instead of random players.


Comment below rating threshold, click here to show it.

Nea De Penserhir

Senior Member

12-10-2012

Summary: Player feels team is holding him back, while not considering the other team is in a similar boat.


Comment below rating threshold, click here to show it.

Flipsy

Senior Member

12-10-2012

Your response to my post shows me you blow at reading. I was stating turrets SHOULDN'T count towards ELO. and in your other thread, someone said that if you went 30-0 farming bottom while the enemy team was pushing down your Nexus, you stated 30 kills was a high contribution, which COMPLETELY undermines your original pretext. You state you wanted your system to be objective-based and not be a fragfest, but yet 30-0 farming bottom is "high contribution" to what? a fragfest? So now you're a liar or have a terrible short term memory.

Your system has very interesting flaws. You stated that deaths wouldn't count in your ELO system so it only includes kills and assists, with deaths being dismissed to allow tanks, who "are meant to die anyway", to die for the team. That's incentive to make it a farming game, or a game full of nothing but turret diving and tunnel vision, which is what low level players do. You also stated that your system is aimed to show the amount of work produced in a game. RamboRal commented saying that "essentially a player who stupidly dove in to try to get kills and die trying will have higher ELO than those that play smart and conservatively" and you said " They don't get rewarded unless they manage to get a kill or assist." First off, that's promoting a fragfest yet again and you contradict yourself, and secondly, I've seen a game go back and forth for 9 minutes before a first blood even happened, and the final score before the Nexus went down was a whopping 10-6. 1 Dragon for either team. Your system would give them less of a bonus than some turret diving **** muffin who sees a low HP Alistar behind a turret and tries to dive to the inhibitor turret for the kill. That's stupid.

The easiest solution to fix ELO, as I've stated before, is simply remove duo queue. ELO is a measurement of your INDIVIDUAL SKILL LEVEL, and should be as such...no friends to duo with you and carry you, and no person for you to duo with and carry. It should also make a timeframe so you don't have the same person on your team as last game to initiate the sense of being "true random". That way you don't get carried again by a good player, and you don't get bogged down again by a bad player. That's it. No contradictory bull**** of "make it objective by forcing players to last hit every Dragon, Baron, and turret, but make it a fragfest by having people with more kills and assists get more ELO" because that's just a mess of loopholes waiting to happen.