Guild Wars 2

I strongly suspect that this is who tends to be queuing for SoloQ, and why the current matcher puts mid level people on very weak teams, holding them back.

The higher your MMR, the more you play. The matcher can easily find people at your MMR level and team you up.

At the bottom, people try it a while and are discouraged. The don’t queue up as often.
They also give up and stand around, sinking the entire team and handing yet another win to already overpowered teams from “above”.

The orange dot is a player at around level 350, getting a team filled from below, not above.

Also, the difference in “win power” between a 900 and a 500 is much greater than between a 500 and a 100…even though it is still a 400-pt difference in rank.

The people at the 900 level are often not serious. The give up and stand around, and a couple of those doing that can completely sink a team.

Thus the win-power of an MMR rank is not linear, it falls exponentially after 500 or so, I bet.

I’m more annoyed that I got a low win rate on day one, and then didn’t play for a while and let rank keep dropping. It went from around 40 to 765 and then I decided to try solo queue again but only play at prime time. I believe I went 6:1 (W:L) and moved up to around 650. I also raised my win rate by about 10% in those games. Since then, not even 24 hours later I am nearly back to the same rank. I was going to try to do 5 or so games every night at prime time, but I have a feeling I will just constantly get pushed back down to the same level. I am not exactly sure why that is. Maybe games played are also factored in some how? I guess if I get my win rate abnormally high, which shouldn’t be possible in a truly random match, it will drop slower and then I can progress at my own rate?

Another thing I find odd is that I had a higher win rate solo queuing in in team pvp than I do as a soloQ player. Logic says that I should have a higher win rate as a soloQ player because I never face premades in soloQ. Anyone care to enlighten me?

Ashenor, you don’t face premades, but you still face much stronger teams once you fall to the level where people your rating do not play as much. You get teammates from below your level. The ones from above are already teamed at their level.

What exactly is this showing us? That players that play more often have a better chance of getting linked up with higher ranked players?

I don’t really understand what your conclusion is from this data.

Silentshoes.1805:

Ashenor, you don’t face premades, but you still face much stronger teams once you fall to the level where people your rating do not play as much. You get teammates from below your level. The ones from above are already teamed at their level.

Kind of disheartening. I guess I will see how it goes tonight at prime time.

Pandabro.8743:

What exactly is this showing us? That players that play more often have a better chance of getting linked up with higher ranked players?

I don’t really understand what your conclusion is from this data.

The idea here is that the matcher seems to first look for people of similar rating to form a team. I suspect that there are more high-MMR people playing soloQ at any given time (because it is fun to win, and they are dedicated players, too). So the matcher easily finds them and tends to put them together on a team.

We have seen screenshots showing this; they should be “dealt” back and forth to opposite teams. But that does not seem to happen.

Once it has some teams, the matcher then seems to pair up the teams by averaging the rating of each team and puts them against each other in a match.

This needs to be seen. This data shows exactly why the current team making system is broken.

Narcin Roc.3018:

This needs to be seen. This data shows exactly why the current team making system is broken.

I don’t have proof of this. But it’s the only thing that really explains what is happening.

People with high MMR ratings get good strong teams, have fun, and play constantly.

Lower MMR, new people “trickle in” and are put up against these waiting high level teams.

If ArenaNet wanted to stop this ASAP, they would issue a hotfix with an algorithm that collects 10 players of similar rating FIRST (not 5), and then tosses them onto two opposing teams.

In the name of science someone should record the results of some matches based on screenshots of the leaderboards and screenshots of the players in their matches. You should be able to determine the rank of each person in the match, and then see if the queue system is doing something like Silentshoes is saying.

Imagine a scenario where 10 people queue with leaderboard ranks ranging from #1 to #100, one person every 10 ranks. Would the match look like this:

#1, #10, #20, #30, #40, #50 vs. #60, #70, #80, #90, #100

OR

#1, #20, #40, #80, #100 vs. #10, #30, #50, #70, #90

Obviously the second scenario should end up in a closer match. I would be surprised if Anet designed their system to intentionally make it so “winners win” by keeping them with other winners.

Either way, someone should be able to figure this out with some time and a lot of hourly screenshots of leaderboards and scoreboards from each game.

Kavia,

Uberkingkong has already posted a number of screenshots with top 20 people all on one team, and significantly lower MMR people on the opposing team. They were what got me thinking about what could explain it.

Silentshoes.1805:

Kavia,

Uberkingkong has already posted a number of screenshots with top 20 people all on one team, and significantly lower MMR people on the opposing team. They were what got me thinking about what could explain it.

I have seen those screenshots, but many were from day 1 of solo queue which is likely a flawed sample. People’s ranks were fluctuating a lot that day and there was a small number of players. Also, none of those screenshots have corresponding leaderboard pics to confirm where the players were on the leaderboards at the time the match was played.

Someone needs to like, take a leaderboard screenshot, play one match, record results, wait for leaderboard update, take another leadboard screenshot, play one match, record results etc. It would be time consuming or I would be volunteering. If a lot of people were to collaborate on it it wouldn’t be too bad though.