LoL needs a role-based queue for efficiency and player behavior (statistical review!)

Comment below rating threshold, click here to show it.

CitizenBeta

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

Junior Member

03-29-2013

My goal for this post is to demonstrate, using statistics, that a role-based queue will:

1. Solve ranked League of Legends ladder inefficiencies (aka ďElo HellĒ)
2. Reduce the quantity of in-game Summonerís Code Violations
3. Be more fun for players


I started playing a lot of ranked games about three months ago. I felt, like many others, that I was winning and losing games in a way that was not reflective of my own performance in the game. Iíve read a lot of opinions about ďElo hellĒ, how to avoid it, how to get past it, etc. This is summation of the advice I found:

1. Have a good attitude
2. Fill roles
3. Dominate your role (thereís no pros stuck in Elo Hell, right?)


Cool. Sounds good. I tried that and I felt like that helped, but do I really have to play like a pro to get out of silver league? Should someone have to play at a plat level to be in gold? Doesnít that sound inefficient to you? But was it? How could I know for sure?
How about some math?

I decided to start collecting statistics. The rules of my experiment were:

1. Have a good attitude and always follow the summonerís code
2. Fill roles (never be the cause of a role dispute)
3. No dodging for any reason. If someone else dodges, it doesnít count.
4. Record the win or loss, whether there was a duplicate role called in champ select, whether there was a dispute over a role in champ select, and whether there were in-game summonerís code violations.


And so I played and recorded 45 solo-queue ranked 5v5 games of League of Legends. My overall win-loss ratio was 25-20. Not bad. After so many games, it didnít really feel role calls had any effect. The math, however, had a different story to tell:

Win Loss Ratio

1. Win-Loss Ratio (total): 25-20 (56% win rate)
2. Win-Loss Ratio (involving no duplicate calls in champ-select): 11-7 (61% win rate)
3. Win-Loss Ratio (involving a duplicate call in champ-select): 14-13 (52% win rate)
4. Win-Loss Ratio (involving a role dispute in champ-select): 4-5 (44% win rate)

In-game Summonerís Code Violations

1. No duplicate calls in champ-select: only 1 game out of 18 games. (5.5%)
2. With duplicate calls in champ-select: 6 in 27 games. (22.2%)
3. With disputes in champ-select: 5 in 9 games (55.5%)


Öwhat?

The data makes it absolutely clear: if you want to have the best possible win-rate, dodge literally every game where two players event mention wanting the same role. Even if they donít argue about it, your odds of winning is lower. And if they argue about it, the game becomes worse than a coin toss, and thereís probably going to be hostile trash talk all game long.

FurthermoreÖ

By being scientific in my adherence to the summonerís code, my results should be skewed by 1/5, meaning that the correlation that I observed regarding disputes should actually be stronger in a blind study. Imagine these same stats but with more at least 1/5 more disputes.

The Fun Factor

The statistics indicate that enough people either donít know or donít care that thereís a pick order in League of Legends to create a hostile environment that affects wins and losses; however, regardless of the ladder inefficiencies, this data should tell us that people are not being allowed to play the way they want to. In 60% of games, at least one person wanted to play a specific role and werenít able to. More to the point, this makes people upset, to the point of having 400% more in-game summonerís code violations compared to games that donít have duplicate calls. On top of that, they care enough to argue with their team about it a third of time, and when they do, in-game summonerís code violations jump by 1,000%.
What could go wrong?

Riot may be adverse to enforcing a particular meta. I donít think this should be a problem since the game is built and balanced around 5 distinct roles. Players can still deviate from the standard meta, but now theyíll have to talk to their team about it from the chair of their chosen role. If anything, I think this could encourage new metas and more actual planning before a game by eliminating the time wasted on jockeying for roles.

Riot also may be concerned about people getting pigeon-holed into playing a single role 100% of the time. I donít think this will be a problem because people can simply choose their role under this system. Right now, if you want to get ranked practice playing top or mid, best of luck, you probably wont get a chance, ever, unless youíre first pick or are willing to risk a dispute. Under a role-based queue, less played roles will have a shorter wait time to get into games. At least you will actually be able to play different roles if you want to.

In conclusion

According to these statistics, with a role-based queue, Riot can:

1. Improve ladder efficiency by at least 20%
2. Eliminate summonerís code violations caused by role disputes, resulting a 10 point drop in overall violations from 15% to 5%.
3. Help players have more fun by getting them the role they want 100% of the time
4. Increase profit! Happy players spend more money.

Cheers!
-Ryan

http://esportslife.com/lol-needs-a-r...ical-analysis/


Comment below rating threshold, click here to show it.

Penny13

Senior Member

03-29-2013

I like your idea and it's not a bad reflection of what the stats probably are although scientifically 45 games are too few, I do understand however that you're a person and those games already took a lot of your time so I'm empathetic to you not wanting to do up to 1,000+ games to make it viable.

So I do appreciate and agree with that you have said and whilst I do agree that a "raid finder" style role selection queue is a good idea there is one large flaw I can see being exploited by trolls.
(also trolls will still troll but nothing can stop that, only bans will minimise it.)

The flaw is e.g.
Top lane is a very popular roll so it could be expected that queue times would be high where as the support queue is more likely to be quite short.
Now let's say I'm a troll who doesn't care about my team, I just want to top and I'm impatient. So I join the support queue and quickly find a game, as soon as I'm in I instalock Tryndamere. Now my team is stuck with a duo top lane and there's nothing they can do about it.

Now having said that WoW uses this system and I have never seen it abused, the vital difference that I see seems to be the "kick" feature in WoW. Everyone knows that although the support/healer queue is far shorter than the DPS queue if they do attempt to abuse the system they will be promptly kicked from the game each time they attempt to do so.

I believe a solution worth looking into is a combination of the two. Queue times may be extended by having to re-queue after each kick but this could well be worth it because:
1. A trade of longer queue times to guarantee a troll free, decent team comp is something I'm confident most summoners would agree to.
2. I believe that despite the initial slog this behaviour would quickly drop off because as in WoW players wouldn't waste time attempting to troll pick when they know they will be promptly kicked.

I believe the problem stems from the lack of evolution in some aspects of the game compared to others ie, the current queue system worked during season 1 when no clear meta was established and roles were far more flexible. Unfortunately as the meta has evolved, due to distinct advantages over a roaming/meta-less play style, the queue system has not and is now far to open to abuse by those who simply wish to do as they please with no respect for team work.

All summoners need to remember that LoL is not a solo game, it is a team game.


Comment below rating threshold, click here to show it.

NecroticBinder

Senior Member

03-29-2013

Quote:
Originally Posted by CitizenBeta View Post
My goal for this post is to demonstrate, using statistics, that a role-based queue will:

1. Solve ranked League of Legends ladder inefficiencies (aka ďElo HellĒ)
2. Reduce the quantity of in-game Summonerís Code Violations
3. Be more fun for players


I started playing a lot of ranked games about three months ago. I felt, like many others, that I was winning and losing games in a way that was not reflective of my own performance in the game. Iíve read a lot of opinions about ďElo hellĒ, how to avoid it, how to get past it, etc. This is summation of the advice I found:

1. Have a good attitude
2. Fill roles
3. Dominate your role (thereís no pros stuck in Elo Hell, right?)
Alright. Having a good attitude is correct. Being able to fill roles, is also good. Being able to dominate a role is also good. However, these are not definitive things, they only lend to victory. Not all roles are equal, and that's the part that matters. A support played in a pure support role is heavily reliant upon their teammates unless they start to diverge from what is expected. As is an ADC to have their team absorb some of the CC and damage.

Quote:
Cool. Sounds good. I tried that and I felt like that helped, but do I really have to play like a pro to get out of silver league? Should someone have to play at a plat level to be in gold? Doesnít that sound inefficient to you? But was it? How could I know for sure?
How about some math?
Well. You are creating an assumption that you have to play at plat level to hit gold. The point the pros make through extreme examples is that not enough games are simply unwinnable to create the supposed elo hell.

Quote:
I decided to start collecting statistics. The rules of my experiment were:

1. Have a good attitude and always follow the summonerís code
2. Fill roles (never be the cause of a role dispute)
3. No dodging for any reason. If someone else dodges, it doesnít count.
4. Record the win or loss, whether there was a duplicate role called in champ select, whether there was a dispute over a role in champ select, and whether there were in-game summonerís code violations.


And so I played and recorded 45 solo-queue ranked 5v5 games of League of Legends. My overall win-loss ratio was 25-20. Not bad. After so many games, it didnít really feel role calls had any effect. The math, however, had a different story to tell:
The rules are fine. Your sample size however tells us very little. It's miniscule! Even if you had 100 games I would be reluctant to trust any conclusions drawn from this.

Quote:
Win Loss Ratio

1. Win-Loss Ratio (total): 25-20 (56% win rate)
2. Win-Loss Ratio (involving no duplicate calls in champ-select): 11-7 (61% win rate)
3. Win-Loss Ratio (involving a duplicate call in champ-select): 14-13 (52% win rate)
4. Win-Loss Ratio (involving a role dispute in champ-select): 4-5 (44% win rate)
This could also imply that summoners who get into role disputes are more likely to fight with their teammates and be aggressive leading to games lost.

Quote:
In-game Summonerís Code Violations

1. No duplicate calls in champ-select: only 1 game out of 18 games. (5.5%)
2. With duplicate calls in champ-select: 6 in 27 games. (22.2%)
3. With disputes in champ-select: 5 in 9 games (55.5%)
Quote:

Öwhat?
This data pool is even worse.

Quote:
The data makes it absolutely clear: if you want to have the best possible win-rate, dodge literally every game where two players event mention wanting the same role. Even if they donít argue about it, your odds of winning is lower. And if they argue about it, the game becomes worse than a coin toss, and thereís probably going to be hostile trash talk all game long.
No. The data makes it absolutely clear that you haven't gathered a large enough pool to draw a valid conclusion.

Quote:
FurthermoreÖ

By being scientific in my adherence to the summonerís code, my results should be skewed by 1/5, meaning that the correlation that I observed regarding disputes should actually be stronger in a blind study. Imagine these same stats but with more at least 1/5 more disputes.

The Fun Factor

The statistics indicate that enough people either donít know or donít care that thereís a pick order in League of Legends to create a hostile environment that affects wins and losses; however, regardless of the ladder inefficiencies, this data should tell us that people are not being allowed to play the way they want to. In 60% of games, at least one person wanted to play a specific role and werenít able to. More to the point, this makes people upset, to the point of having 400% more in-game summonerís code violations compared to games that donít have duplicate calls. On top of that, they care enough to argue with their team about it a third of time, and when they do, in-game summonerís code violations jump by 1,000%.
What could go wrong?
The problem with this is that it enforces a meta. That's what is guaranteed to go wrong by nature of the system. The game is designed to deal with five champions versus five champions.

Quote:
Riot may be adverse to enforcing a particular meta. I donít think this should be a problem since the game is built and balanced around 5 distinct roles. Players can still deviate from the standard meta, but now theyíll have to talk to their team about it from the chair of their chosen role. If anything, I think this could encourage new metas and more actual planning before a game by eliminating the time wasted on jockeying for roles.
See above. And no, it really wouldn't. Players are hostile enough in regards to people violating the meta when there isn't an official way for it to be organized. By creating a system that reinforces this, people are going to be more hostile to those who break it.

Quote:
Riot also may be concerned about people getting pigeon-holed into playing a single role 100% of the time. I donít think this will be a problem because people can simply choose their role under this system. Right now, if you want to get ranked practice playing top or mid, best of luck, you probably wont get a chance, ever, unless youíre first pick or are willing to risk a dispute. Under a role-based queue, less played roles will have a shorter wait time to get into games. At least you will actually be able to play different roles if you want to.

In conclusion

According to these statistics, with a role-based queue, Riot can:

1. Improve ladder efficiency by at least 20%
2. Eliminate summonerís code violations caused by role disputes, resulting a 10 point drop in overall violations from 15% to 5%.
3. Help players have more fun by getting them the role they want 100% of the time
4. Increase profit! Happy players spend more money.

Cheers!
-Ryan

http://esportslife.com/lol-needs-a-r...ical-analysis/
In conclusion. Your statistics are far to small to be taken seriously. You fail to understand the psychology behind players that already exists that would be reinforced by our system and have made quite a few false assumptions and false conclusions.


Comment below rating threshold, click here to show it.

Penny13

Senior Member

03-29-2013

The meta has been formed through trial and error, granted amongst professionals, because it has so far proven to be the most viable and competitive way to play LoL due to roll and gold/exp distribution, there is still a lot of flexibility within the meta itself. Lane swapping is a very simple example.

I don't believe that an enforced meta would be overly detrimental or opposed. WoW for example, although a very different game has a strictly imposed meta because it works so well and due to this players are more than willing to fill and comply with roles.


Comment below rating threshold, click here to show it.

NecroticBinder

Senior Member

03-29-2013

Quote:
Originally Posted by Penny13 View Post
The meta has been formed through trial and error, granted amongst professionals, because it has so far proven to be the most viable and competitive way to play LoL due to roll and gold/exp distribution, there is still a lot of flexibility within the meta itself. Lane swapping is a very simple example.

I don't believe that an enforced meta would be overly detrimental or opposed. WoW for example, although a very different game has a strictly imposed meta because it works so well and due to this players are more than willing to fill and comply with roles.
Because WoW's dungeons are designed with the meta in mind. And what happens when people decide to flat out change the meta? Besides, how do you even determine what the roles ARE? Would you be signing up for an AP top lane? Or would you be signing up for the role of bruiser, etc.

The suggestion of implementing a role queue has been brought up time and time again and will be shot down time and time again. It just hurts the game.


Comment below rating threshold, click here to show it.

Lagenor

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

Senior Member

03-30-2013

Quote:
Should someone have to play at a plat level to be in gold?
That's rediculous. By this logic, all gold players are plat players. So all gold players are silver players, so all silver players are gold players.

No

Tiers skill level change over time. That's part of a growing game. Do you really think the top 100 players right now are just as skilled as the top 100 players 2 years ago? Of course not.

ELO Hell is an illusion, it just does not exists. Everyone get affected by "feeders," AFKrs, etc. at what ever tier they're having that problem in. ELO Hell = what ever tier you're trying to get out of. Because you always find faults in others. "OMFG This twitch is feeding! How am I paired with such trolls!" In reality, you climb the ladder by being that 1 player in that equally skilled match is improving. By doing to you will be that advantage that your team constantly has just by having you. You alone makes your team win 5-10% more means you WILL get out of your ladder.

I would actually hate ranked more if they provided a role system. Trolls would abuse it more ("I want to ADC as amumu!, because I want to play with my DUO partner and not listen to the meta!") People would complain just as much as they do now. The new "ELO Hell" would be exactly what it is now, but for similar reasons.

I honestly feel that 45 games is not a good enough sample. If you're at your ELO Hell, whichever division that is, bronze I, silver II, whatever you consider it to be, It will take you more than 10 games to climb out. Sometimes it takes more than being in the promotional series 2 or even 5 times to get out, because you will ONLY get out if YOU improve your skill more than what it ALREADY IS.

You will get to gold, when you play at gold. Not platinum, because, again, by that logic, in order to hit platinum, you would have to play at diamond tier, and to hit diamond, you have to play at challenger tier. So to hit challenger tier, you obviously must be the best player in the world. But oh wait, there is more than 1 of those in challenger.

tl;dr, ELO hell is an illusion, you will get to gold when you play at a gold level.


Comment below rating threshold, click here to show it.

YouTastedPurple

Member

03-31-2013

well you know 1 pearson will neva be able to do the 1k to make this sort of test legit well how about we try and trust people and do a servay each player dose 10 games post their findings under your rules with a screanshot of their match history 10 players do this hay look at that 100

just might be a better way of doing this

things to take note of would be

1. Win-Loss Ratio (total)
2. Win-Loss Ratio (involving no duplicate calls in champ-select)
3. Win-Loss Ratio (involving a duplicate call in champ-select)
4. Win-Loss Ratio (involving a role dispute in champ-select)
5. rolls in dispute
6. your roll
7. if this is your optimal roll
8. if there was something not in the meta and weather there was any disput about this


i dont know just throwing this out there id say make a proper list of these thing get as much data as you can and just hope you dont get trolled for trying this

and even if its not realistic for this to happen it would be nice to be able to say hey this is 1k or even 2k games and this is my result

if you had a comprehensive servay riot might say hay we could try this in beta for x time just to see if these results change


anyway rant over and i will help next time im on if you have a list of rules to follow


Comment below rating threshold, click here to show it.

CitizenBeta

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

Junior Member

04-03-2013

Without individually responding to every point, here's what I gathered as the main critical responses:

1. Sample size is too small.

It's really not. With a sample size of 45 games with that strong of a correlation, it's statistically viable.

2. "You are creating an assumption that you have to play at plat level to hit gold."

That's not what I'm saying at all. What I'm saying is that without a role-based queue, a percentage of games are won and lost according to how lucky you are in terms of avoiding duplicate calls and role disputes. This means that the system is slower at accurately placing you at the proper elo than if it could reduce or eliminate duplicate role calls and disputes.