Lyte is amazing, he got back to my e-mail in less than 5 mins!

I mean, duo queue may or may not be an issue. I find that when I duo with someone 100+ elo higher than me, and we go bot, our solo lanes get **** on because they are up against someone 100 elo higher than themselves. I think part of the problem lies in the fact that 100 elo, say, from 1200-1300 is vastly different than 1700-1800. There is some giant gap every 100 elo at the 1500-1900 level in terms of consistent play that makes it difficult to use the same duo queue algorithm across all elos. I don't know how to fix this, to be honest.

And these numbers still don't represent what won a game. A game can be won by one good taunt at the right time, independant of any scoring.

My point is you can't truely rate what won a game by looking at the stats, which is why there are no performance based ratings.

Kills dont neccesarily lead to wins though. That would encourage me to make a super split push strategy, get minimal kills, (even more deaths,) and just push to win without many kills for super elo gain. Just saying that this might be a legitimate situation

i simply assumed this was for ranked queues... if not then ignore this but I only expect to play a minimum of 200 games b4 reaching my true elo because in both seasons I was trolled in half my placement matches. I've climbed from 1100 to 1600 and only cause of those **** trolls in placement matches. If only having someone on your team 1-12-2 or it being (<5)v5 (afking in fountain and only moving once every 2 mins counts) means you lose less elo that game.

I wrote this over a year ago.

http://na.leagueoflegends.com/board/...d.php?t=385895

Explains pretty well why pure elo doesn't work.

conclusion:
Elo, predicted win rate: 54.97%
Margin of Victory, win rate: 63.03%
Actual win rate: 64.3%

I don't think Duo should be an option for ranked personally.

If you want your solo queue Elo, go solo.

This will prevent a high elo smurf from carrying a friend, or large differences in elos when a 900 and 1700 queue.

I think there is also a certain bias where people think that their true elo is the highest elo they've ever achieved, when that was probably the highest elo they've gotten lucky enough to achieve through good teammates carrying them, as evidenced by "being trolled down" once they hit that apex.

Part of the problem is that you can't realisticly rate a duo q with some kind of calculated elo adjustment. In the end this doesn't really matter, it's just another random factor in the matchmaking pool.

This is correct, I almost always win with Ashe but have a poor KDA just because a well timed arrow ends the game after the 40 minute mark.

It's good to see a red post like this.

I have 4 accounts with different ELO for ranked game. However, I have similar win/lose ratio for normal game.

The problem is quite obvious here right?

I believe in ranked game, there should be some condition that the system re-activate one player's basket switch, and change his elo greatly to find his true position.

For example, in a 900 ELO game, one player whose elo basket is determined got 20/2/19 or something like that, and it happens 3 times in the last 5 matches. Then in the 6th game, if he win, he should get 50 elo rather than 10. This burst should be removed when he is not able to make such great carry for a number of matches.

I feel like something should be done about leavers and AFKers. For example, if someone leaves the game they should lose double ELO. Or something like that.. Hmm now that I think about it.. There are different levels for leavers already in place, right? How about people who leave games often are hit with double ELO loss.

Either way, I don't think it's fair that the people on a leavers team should suffer the consequences of a loss.

Make sure Normal Draft mode isn't giving priority to pre-made groups please! Randomize everyone!

Can you give oyur thoughts on This idea i came up with? http://na.leagueoflegends.com/board/...3#post21027543

I agree with you that players who intentionally rage on someone with the goal of making them quit will eventually land in the Tribunal and be punished; however, we have now created a system where players know it is possible to escape from Elo loss if someone leaves. What does this do to player behaviors?

1) The toxic players will rage and force someone to quit. They will be banned by the Tribunal; however, maybe 1-3+ players quit because of their horrible experience hearing "uninstall the game, you suck, never play again." We lost 1-3+ good people, and still had to ban the 1 toxic player.

2) Players who are moderately toxic now know they have a way to force Elo loss on someone else. Before this feature, these players would have probably gotten frustrated but just played another game or taken a break. Now, they might decide to instead rage on who they believe is the person responsible for the loss. Now, we have made a toxic player more toxic than before, and lost still 1-3+ good people.

Even if it is just 1 game, even if the toxic player is banned, I lose sleep at night thinking about the 1-3+ good players who had a miserable, miserable experience.

This is not to say it is a terrible idea; but there has got to be better solutions to consider first.

I think the most frustrating thing about ranked queues is simply that regaurdless of any previous experience you can be queued with people who just hit 30 and have yet to learn all the things that lol has to teach you , i.e lasthitting, warding, gaurding jungle etc cant really blame them as not everyone is diehard enough to watch streams and some just want to play but its frustrating trying to keep it in check when players instalock 3 ranged carries and call first time ranked, obviously not all new 30s are like this and obviously there are trolls out there who reach 30 just so they can crash and burn another ranked profile but to get out of the default starting elo to places where people have at least an inkling of whats going on is just dice rolling to see who gets the worst team obviously basing ranked stats on normal games but is it really ideal to simply keep playing ranked for 200 or so games trying to filter losses trolls and newbies out(not noobs as this is an insult and i dont feel new players should be insulted for not dedicating their lives to LoL). Also reporting trolls at the end of the game doesnt really help cause their damage is done. - if ive sidestepped your point im sorry, this has been bothering me :/.

I want to focus my attention this thread for now, but I've bookmarked this for later reading this weekend.

Something needs to be done because it's not fair at all to the other 4 players that didn't rage quit.

You can't value current rankings as the acutal true rankings. A player has a range where he belongs to not a set elo value. This is due to the randomness of the system. The great thing about all of this is that it doesn't even matter if the players rankings are correct, because over a large set of matches errors/misplacement/other random factors cancel each other out, which makes your ranking more and more a rating of your actual performance compared to the other players perfermances, it's basicly a comparison. That's the beauty of it and it results in a Normal Distribution.

People need to stop looking at single games, and start understanding that the system measures your overall performance over all the games you have done, compared to that of the other players (and all the games they have done).

I found your solution. If one or more people bully someone else into leaving the game the leaver isn't punished outside the normal leave/Elo loss and the people who bullied the person to leave get perma banned instantly without question and cannot recover that account.

Huh. Problem solved. Ranked gets fixed and no bullies.

This is an interesting idea; however, what is to stop a team from stalling a game so they can pad their Team KDA and continue stomping for higher Elo gains? The other team should surrender; however, what if there is one or two players on the other team who refuse to surrender and choose to now troll their own team?

There should be no "Duo" queue in a "Solo" ranked game, that takes all the skill out of carrying and puts the game in your favor, especially if you have two very good players playing together, it just becomes completely unfair. The level of communication that 2 players on skype or vent have is immense compared to those typing in game. A duo lane bot support and a carry becomes very easy to communicate with a service like skype, or perhaps one person in the duo goes top or mid and the other jungles constantly ganking that lane because they can communicate exceptionally well. The win potential that is given to such pairs is unfair because it is not an accurate representation of your own skill in a solo ranked game.

Added thread to my weekend reading.

This.

Especially the trolls that intentionally feed.

I still wish you guys would change the system entirely for ranked. I know it can work for some team games, apparently in College Football(the bcs) it is used to some reason, but that is still 60 people or so on one team in one place who play together.

In solo queue(even duo) it is just random people who may or may not communicate. A system originally designed for a 1v1 game(Chess) and that can work for Team games, where the team actually plays together more than that 20 minutes to an hour(or so) that a game takes, you can't get a true reading out of it.

I had a suggestion a while ago, that I can't find anymore(I tried the last 500 post thing, but did not find that one)that would kind of normalize it where the elo gain and loss was based on several factors. From K/D/A(not important overall) to minions, to wards, and several other things that could allow someone who was playing well, and just on a team that COULD NOT WIN ever, in a million years to not just lose that 10-60 elo for the match, and also would hurt someone that the team carried to not gain the 10-60 elo immediately, they would have to have helped in some way to earn the full amount.

It basically broke down that the winning team could have people that gained 0 elo, and the losing team could have people that lost 0 elo. It was a good idea, but sadly has disappeared and it will take a bit for me to rememebr it all.

You could give the adjustments limits to prevent this. But regardless, across all sports and ranking systems, margin of victory massively improves ranking algorithms. No competent sports ranker, or even someone who sells picks, will simply use wins and losses. Or even opponent adjusted wins and losses.

Well, this is coming from a non-ranked scrub, but I believe duo queueing is a major culprit at hand, much like the others have said. And again, I feel that the "You can't play with your friends!" idea is false, as there are always normals and customs where you can train and help your friend just as much, without the additional frustration you may provide other players in ranked due to the significantly worse player you brought along.

We have talked about this idea in-detail before, but it is not necessarily in the scope of this thread. I might do a dialogue about this specific issue in the future.

Can I ask what Riot thinks of their current Dominion Elo? My EUW account (Kittal) is ~1600 Dominion Elo, but I recently had a game where I was against two people on the enemy team at 1700+, and three at 1550+, and everyone on my team was 1400-1500 Elo, according to Jabebot which I imagine is reliable. It felt like a really unbalanced match - not just from looking at the Elos, but it was evident by the way we were stomped in game.

Yeah i agree completly but there should be an option open to those 1-3 players to get rid of said toxic player right there and then , how many times have we all been solo qued with someone raging their face off and have to put up with it for 40 minutes, maybe an ingame reporting system complete with a message in chat would serve as more of an immediate stop what you are doing warning to that player then a possible unnoticable report after the game has ended, if the tribunals as fair as we like to believe it will deal with them if they continue to break what ever rule they are breaking and show leninacy if they stop once ingame warned , after all if they know its gonna go to tribunal whos gonna keep digging their grave, only problem I see is what ever tools are given to players are also given to toxic players and are open to abuse by 4 man premades

Why not have ELO based on performance? So if you play well on the winning team, you will gain a little more ELO. If you do well on a losing team, you will lose a little less ELO. If you do poorly on the winning or losing side, you gain or lose the normal amount of ELO.

I will play a few games this week of ranked soloque + douque ranked + solo normal + full party normal + ranked teams and email results to you with anything i noticed. ill try to do this by the weekend.

but what i have noticed is that starting ranked anything isnt really "fair" and for most people its "scary".

the reason behind this is up untill a player reaches 20 he has not been exposed properly to how ranked games work, what the stakes are at and much more. they dont learn proper warding while leveling because you are more focus on gaining exp and rune pages / more champions/

it might be handy to have a minimum normal draft game requirement before playing ranked.

in my personal experiance of starting ranked for the first time, i admit i was clueless as hell. i went straight into it after only 150 wins (at almost 500 now didnt play for half of season 1) so i was que'd up with players who knew alot more about the game then i did at first. resulting in a waste of my placement games. i still ended up in the 1100 range. however i didnt learn anything useful to advance. it took me quite some time of watching streams and my own replays and looking up reason of what to do and when before i actually felt like i can go into soloque and benefit my team positvily.

however due to this and it happening to so many people. it does not give accurate elos to anyone unless they had pure luck or they played broken champions. in some of the lessons pros give they say the easiest way to get out of the low elo's is to play the currently broken/ over powered champions. champions that take good knowledge of the gameplay and how to counter pick/ ward/ and more to fight against.

there needs to be a way to teach the new people coming into the game on how to make smart choices. there also should be a better grace period for the ranked games judgement period. maybe instead of shifting elo brackets after every win or loss at first you can shift to a average of 5 games in a certain bracket style then average it out play 5 games in that new bracket and go up or down to adjust for a decent amount of games before they are ranked.

No screenshot, but let me say this. Someone with 50 wins and 35 losses who is at 1750 ELO is not a true testament of their skill in Dominion. Almost every game I play, if I don't get 20+ kills and carry like a boss, it's a loss. I can't count the number of games where I get 1st place, control points like crazy, yet I always have this one guy who "lucked" his way in to a decent ELO, yet knows nothing about Dominion. I'm sorry, but if I can carry 2k ELO games, why am I getting stuck with people who don't even know how to play and cost me tons and tons of ELO, resulting in me being endlessly grouped with bad players. Even when I group with my 1900+ ELO friends, it never seems to fail that I get 1st place, have a huge score and they have bad games. Maybe it's just bad luck, but I truly and tired of being grouped with people who have very few (less than 100) wins all the time and proceed to troll, fail or just don't take the game mode serious.

Lyte is becoming my favorite Rioter at this rate. Thank you so much for taking the time to talk about this with us. Nothing would make me happier than having intense, nail-biting matches on a permanent basis (because let's face it, stomping isn't too fun after a while either).

They can still play together in normals. Ranked should be a place for players to gauge and prove their strength in a measurable way. As it is effectively impossible to match people by their preferred role, duo queueing when people are far apart will inevitably create match ups like a 1700 vs a 1300 top in which case neither player really learns anything. The game may still be balanced because there is an equally bad match up somewhere else, but it still feels like a broken mechanic.

Separate tweaks for the Elo systems for Dominion and Twisted Treeline is something I am interested in; I need more data, but my gut is that there could be some independent tweaks to just Dominion or Twisted Treeline Elo systems that can help.

The only thing that could be done realisticly is some elo bonus or mitigate some of the losses in very extrem and obvious cases (2v5, 3v5). I still don't see how this changes anything though. It still happens to both teams about 50% of the time, so this like everything else just evens out again. It's alot of work for Riot that doesn't change anything.

Even with the perfect system, most of us would still be at the same ranking at the end. The difference is some just haven't realised this yet.

Somebody give this man a raise!!!!

Constant lack of supply of the support role. 10 hour queues.
Forcing metagame onto playerbase, destroys creativity with new metas/roles/plays.

yes. I want kill lanes bots or gooose karthus/soraka bot