Q:
Is MMR better than being truly random?
I’m really curious. Obviously ANet believes so or else they wouldn’t implent one but I look at the forums every day at work and I can’t help but wonder if all the negative feedback would be muted if solo queue was random queue.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Personally I feel like Ranked should be more akin to a random seeded tournament and if you get knocked out by the tournament winner first round… At least you can blame the dice not the invisible hand.
I’d like to know the answer to this question too. I recently posted about this topic and didn’t have any serious posts. In my opinion, a system where 10 random players, in the same division, tier, pip range, are randomly assigned to blue or red team would be the best system. As an example to give you guys an idea, place 10 players names in a hat, shuffle hat, pick one name from the hat and that person is on blue team. Pick another name and that person is on red team. Continue until all names from the hat have been pulled. Seriously, this is undoubtedly the best system. But of course all the “MLG” try hards will wine and say their games are inconsistent. And, also many people will get a reality check that they arnt as good as they think they are.
MMR works (or at least used to work) to balance matchmaking so that matches are fair. In season 1, this worked to make matches generally enjoyable because the level of competition was more likely to match your own. The problem with this was that it’s not competitive and completely contrary to the term “ranking”. Your rank was more likely determined by how many games you played, not how well you played them.
In season 2, they maintained the same system, but disabled comparison between teams. They claimed that matchmaking would be based on tier (points) but that initially we would be matched by our previous MMR (implying MMR would depreciate?) but it never did. The intention for S2 was to disable players with low skill and accelerate players with high skill. The problem however is that by strokes of luck during kickoff, MMR became biased and unrecoverable, regardless of whether you were on the winning or losing side, and locked many good players into an either guaranteed permanent win or loss streak. This is why we have people coming from ranked with 3.0 w/l ratios now having 0.3, and absolute scrublords in diamond after only a few days. These random distributions will inhibit all true competitive play until all players are equally spread out across the tier fields. However, we’re talking hundreds of repeats in the legendary division to achieve that dynamic.
A proper ranking system would have worked in only one of two ways:
1) Rank purely based on MMR where everyone starts at zero. The problem with this is it the reward system has to be changed. You can’t award pips by wins, but instead have to award them through MMR progression. That means no win streaks, loss rebounds, or safety nets. Pips have to represent MMR in a static manor. Another major benefit to this method is that you can constrict pips to a specific range (ending at legendary).
2) Scrap MMR entirely and base all matchmaking on pip tiers alone. The benefit of this method is that not much has to be changed. MMR and pip loss safety nets would have to be cut, including the protection against division demotion.
S1 and S2 attempted to mix both of these at the same time and it doesn’t work. The two reward systems will fight each other as soon as one passes the other, causing heavily biased matchmaking that either inhibits all progression (S1 nullification), or forces full progression in a biased direction (S2 perm win/loss streaks).
(…)
The problem however is that by strokes of luck during kickoff, MMR became biased and unrecoverable, regardless of whether you were on the winning or losing side, and locked many good players into an either guaranteed permanent win or loss streak. This is why we have people coming from ranked with 3.0 w/l ratios now having 0.3, and absolute scrublords in diamond after only a few days. These random distributions will inhibit all true competitive play until all players are equally spread out across the tier fields. However, we’re talking hundreds of repeats in the legendary division to achieve that dynamic.
(…)
I noticed this happening on my 2nd account yesterday, and thats even though season start should be over now… and I was waxing about it in another thread… about how many matches it will probably take to re-adjust that guys mmr in ruby then.
So this really is a thing, its not just some crazy idea of players stuck at their limit.
I noticed this happening on my 2nd account yesterday, and thats even though season start should be over now… and I was waxing about it in another thread… about how many matches it will probably take to re-adjust that guys mmr in ruby then.
So this really is a thing, its not just some crazy idea of players stuck at their limit.
No it doesn’t place anyone “where they belong”. If you want to get where you belong as a team, you can get there with 5 man guild teams, just not as a soloQ.
From the recent games in high diamond. It looks like mmr is a non-existant thing. Everything from people going boss at start of forest to pushing lords with 2 people at 110-90.
I don’t think mmr should be utilized together with Leagues. I think League placement should be the only rating they count.
Sure, it will be a complete kittenshow at season restart. But then again, in what game isn’t there?
Optimise [OP]
From the recent games in high diamond. It looks like mmr is a non-existant thing. Everything from people going boss at start of forest to pushing lords with 2 people at 110-90.
I don’t think mmr should be utilized together with Leagues. I think League placement should be the only rating they count.
Sure, it will be a complete kittenshow at season restart. But then again, in what game isn’t there?
Lol@high diamond tiers going for beasts at start in forest. Dem skilled players.
BTW, I agree with scrapping completely the MMR from the equation, get 10 players inside a pip range, and distribute them randomly between the teams.
OR, the other solution, but way more drastic: make ranked like GW1 GvG – only full 5 players premades can enter it. Leave Unranked for soloqueuers. This is a kitten team game. But I guess this would go against the majority of the ego-inflated individualist player base.
2) Scrap MMR entirely and base all matchmaking on pip tiers alone. The benefit of this method is that not much has to be changed. MMR and pip loss safety nets would have to be cut, including the protection against division demotion.
I would prefer ranking purely by MMR, but this seems more likely to be the system Arenanet would implement, if any. Removing safety nets is essential though, and I’m not sure if they would ever do that. Also, the first week or two would be a crapfest similar in magnitude to the start of season 2, but perhaps that could be alleviated with a pre-season.
Rightful question.
Answer is obvious – no.
They matchmaking system is worse than pure random selection.