Another attempt to get the attention of Anet. Maybe some of the things are mentioning are already the case.
For the matchmaking, some things that were added to the new system seem to be backfiring. They added a new variable to avoid teams containing multiple of the same class. This was done by upping the “MMR” of the team by a certain amount depending on how many duplicates or triplicates they have. However this isn’t fully working because 2 teams containing 2 of a certain class will even eachother out. Can’t really say how what i’m about to suggest can be properly coded. But the better system would be to reserve slots per class in the pool. Currently the queue is supposedly collecting 20 peeps to slot into 2 teams. What if the queue was set to collect 3 of each class before matching into 2 teams. Take into account not to deplete a stack of classes rapidly though if it slots less (think impopular classes). And don’t use duplicates/triplicates as a means to up the MMR score of a team.
I’m asking to avoid this, is because a problem with triplicates would have, is they’re facing a team that is actually better than them, because the other team averages 100 MMR more than them, because the duplicate/triplicate artificially made em seem stronger. This is also an issue for groups of mediocre players who aren’t using TS to boost their performance. The matchmaking system is upping them to higher level than they truely are (also the reason why top premade teams have to wait a crapload of minutes of games, they’re top of the MMR, and the fact they premade ups them even higher).
Now to MMR ranking itself. There’s a big problem that the “pack” of PvPers are often being put with newbies in their teams. This isn’t fun for the newbies cause they’re getting stomped left and right, nor for the vets who keep getting them in their teams because they’re technically pushing them in a 4v5 (or even worse if they’re being rally bots).
As far as i can see the system of MMR is that a certain score is given to the winners and that certain score is taken from the losers. What Justin once mentioned in a topic is true, just upping the start score of people won’t help this issue as the median would raise along. This is my suggestion:
*Lower the start score, say from 1500 down to 1000-1200.
This is to seperate the new players from the veterans. How to split the start score from the median will follow further on.
*Lower the starting variance, raise the final variance.
This will give new players still a quick grow or decrease in MMR, but less extreme then now, that way lucky starters don’t get instantly stuffed amongst veterans. And with a higher variance in the end, people who once got carried by a great team, but suddenly fare less succesfully will move swiftly to the “pack”.
*While below 1500 MMR, you gain more rank points than you take from your enemies. If you lose you lose less points than your enemies. See the following formula done on the MMR gain/loss based on the current calculations.
actual pt gained = mmr gain * 1500 / your rating.
actual pt lost = mmr loss * your rating / 1500.
*To avoid that too much rating is being fed into the ladder by baddies, do the reverse at higher scores. like this for eg above 2000 rating.
actual pt gained = mmr gain * 2000 / your rating
actual pt lost = mmr loss * your rating / 2000.
Due to the fact there are more baddies than there are great players, you’d be creating more points in the 1500 range than you’d be taking away from 2000+ until more and more are pumped into 2000+ MMR, but baddies will most likely still remain under the 1500. The only difference is the median will be pumped, because 1500-2000 will be the heaviest populated. Baddies will be lingering closer to where they started 1100-1300. This way you split the started from the pack as well.