All potential "fixes" for unbalanced premades are fundamentally flawed.

BluePolarizer·Jun 1, 2016·7 votes·501 views

There are two ways to handle unbalanced premades - because they will happen.

do nothing about it. Cool, enjoy getting stomped because their mid/jg/bot were coordinated while you had a lone ADC.
increase the MMR of the premade with lower counts.

This one I have a fundamental problem with.

Let's say that you are playing soccer.

On one side of the field is a high school team that played together for years.

The other side is random players.

If the random players were also high schoolers, they'd get destroyed.

The "MMR compensation" is basically giving the random players, a random guy from Real Madrid, and calling it balanced.

As you see, there is no way for uneven premades to be truly evenly matched.

If they cannot even be evenly matched, how do you compare them on the same ladder?

12 Comments

BloodyTomponJun 1, 20162 votes

Did you know that league has had offsets in MMR for years? It was so unbalanced that you didn't even notice. :P

BluePolarizerJun 1, 20163 votes

I recognize it was there for duos and it was noticable. In fact there's also an MMR offset for red vs. blue side, and it is unbalanced. It's like the left side of a soccer field has a bigger goal, so they allow college players to play on the left side in a high school game.

BloodyTomponJun 1, 20161 votes

Red vs blue had almost exactly 50% win/loss. That's the very definition of it being balanced

BluePolarizerJun 1, 20161 votes

it should be exactly 50% but it isn't, which is why there is an MMR offset. There was a statistical analysis on this a few pages back.

In addition, even with an MMR offset such that the win rate for premades vs. solos is 50/50, it still doesn't solve the problem of rank integrity where the MMR of the premades and the solos measure the same set of skills.

BloodyTomponJun 1, 20162 votes

After the offset, it is 50%. And again it is the definition of balance unless you construct some tautological/definitional truth of anything with an offset is unbalanced.

GaleWinUnleashedJun 1, 20161 votes

It's not. That balance is achieved by giving red side slightly better players. There was a recent in-depth statistics post by a player that ran analyses on over 40 thousand games.

BloodyTomponJun 1, 20161 votes

After the offset it's 50%

GaleWinUnleashedJun 1, 20161 votes

It means red side wins because they have better players, and blue side wins because they have the camera angle advantage.

The 50/50 winrate statistics aren't lying, but they aren't being presented in a fair way either. Balance would be better achieved by allowing the camera to be flipped 180 degrees at the player's discretion. Problem is that this is apparently difficult for Riot to implement, and also comes with some disadvantages, so it's easier to just continue tipping the MMR balance in favor of red side.

BloodyTomponJun 2, 20161 votes

Balance is the 50% win rate (in this context at least: Champ balance is a whole different ball game). It's an output driven goal. The how is not that important.

What (I think) you're talking about is fairness. Fairness is mainly about perception and closely associated with match quality (still subtlety different).

I imagine that there are very few people who feel (in an emotional sense) as if the camera angle is unfair. Few people get hit by a skillshot and say to themselves 'that was so unfair-The camera angle fucked me over there.' Instead the feel the Nid spear is unfair. I think this map offset is almost completely a cognitive unfairness rather than a visceral one. Generally its the visceral feeling of unfairness that is problematic, I don't think we have that.

In terms of offsets for pre-mades, it's only an issue if it impacts on game quality IMO. Riot might care if the game quality is good but people think it's unfair, I highly doubt that though. As a tangent I see it as more likely that they take it in their stride as people wanting to blame external factors. I suspect Riot actually likes it. If people are blaming the pre-mades they are not blaming their team. Additionally it creates a kind of 'bitching bonding moment' - Have you ever bonded over a teacher the entire class hated? I see it as a marginal gain in both reducing toxicity and improving player bonding. End of tangent :)

PrometheSJun 1, 20161 votes

Last time I post this but:

Clumping every queue configuration together makes no sense for a ranked ladder since solo to 5-man are different skills and experiences (plus possible voice chat/pre-game advantages). No matter how hard you try to adjust it, it's inherently a broken system if you take competitive play seriously. There is simply no other way around it. That's the problem for me. It's not just unfair matchmaking, it's that it lacks validity as a ranking system since it doesn't force consistency in the experience despite putting everything on a single ladder. The rankings are too ambiguous now due to allowing every queue configuration and are no longer very reflective of either 5 man or solo performance. It's just some weird middleground that doesn't do a good job at anything.

They could at least show a solo ranking separately though even if it's all DQ still.

BluePolarizerJun 1, 20161 votes

completely agree but many people don't like this sort of suggestions for some reason.

Perhaps they have disproportionately benefitted from dynamic queue and lack of transparency on who was solo and who wasn't.

PrometheSJun 1, 20161 votes

{quoted}

completely agree but many people don't like this sort of suggestions for some reason.

Perhaps they have disproportionately benefitted from dynamic queue and lack of transparency on who was solo and who wasn't.

It wouldn't have to be retroactive anyway. It could just be from whenever they introduce it forward. While it would still be pretty shitty for solo players, it would make it way more predictive of individual skill. Sometimes you'll get carried by pre-mades, other times you'll get destroyed by them but the variables will be pretty much the same for all solo players over sufficient games. Thus, the ranking shown would be about as accurate as the old solo/duo system (which was also slightly distorted). Implementing this would not harm queue times or do anything else negative. Unless Riot is actually intending to obfuscate individual skill, there is no downside.