66 replies to this topic

[DeathlyEyes]

Reduce the number of players to make individual performance more important. This should make more balanced matchmaking.

[Mawai]

InRev, on 06 April 2015 - 03:31 PM, said:

9 people on his team didn't even break 100 damage. That includes 2 of his team's assaults, which were also hero/Loyalty mechs, implying the players should have at least a passing grasp of a clue.

In other words, his team got all the donkeys, which is a bad MM failure.

Nah ... probably the folks on his team ran off to engage the opponents ... leaving the assaults behind while an opposing lance or two came across these two assaults on their own, flanked them and mowed them down. The 4 to 6 mechs in this group then proceeded to roll up the defensive line of the opposing team which was facing the wrong way. Wiped the team 12:2 due to lack of team work ... not lack of players.

Honestly, in most of the matches that I have been that wiped it is usually a scenario like that I described above that loses the match ... nothing to do with players carrying harder, nothing to do with matchmaker fail ... just not playing as a team.

Also, to be honest, I have seen lots of decent players make donkey moves ... survive some and die to others.

[Sir Wulfrick]

I've been keeping end-of-game screen shots recently which reveal some interesting statistics:

Over the past week or so I've played 82 games. Of those games only 12 have resulted in one team's deaths being within 50% of the other teams' deaths, i.e. 70 out of 82 games have been near-total wipes in one direction or the other.

85% severely unbalanced game results cannot mean that the match maker is working as intended. Granted, 82 games is only a tiny sample size but I think it's enough to reveal that clearly there's a problem. This data doesn't say what that problem might be, but very obviously there is one.

[Novakaine]

Sir Wulfrick, on 07 April 2015 - 05:07 PM, said:

I've been keeping end-of-game screen shots recently which reveal some interesting statistics:

Over the past week or so I've played 82 games. Of those games only 12 have resulted in one team's deaths being within 50% of the other teams' deaths, i.e. 70 out of 82 games have been near-total wipes in one direction or the other.

85% severely unbalanced game results cannot mean that the match maker is working as intended. Granted, 82 games is only a tiny sample size but I think it's enough to reveal that clearly there's a problem. This data doesn't say what that problem might be, but very obviously there is one.

[RazorbeastFXK3]

As far as I see it.. the matchmaking works according to what weight 'mech you bring to the queue and what rooms are available with positions for your choice of tonnage. Some days if you choose a light 'mech you'll be thrown into battle faster than if you were in an assault. Then there are times if you're in a medium 'mech you'll get thrown into battle faster than if you were in a light or a heavy.

There are times when I'll wait a full 3 minutes to get into battle then there are times when I'll get thrown into battle within 3 seconds of me hitting the 'Ready' button. One time I thought I crashed 'cause it blackscreened within one second of me clicking the button to find a battle to join.. nope.. I was just thrown into a room that was waiting for me.

So if anyone asked me.. I believe the matchmaker program that pairs you up with players who match your stats doesn't even exist.

Crazy Billy Joe Bob, on 07 April 2015 - 10:54 AM, said:

nobody understands how how the match making is done, not even PGI.

[Joseph Mallan]

Sir Wulfrick, on 07 April 2015 - 05:07 PM, said:

I've been keeping end-of-game screen shots recently which reveal some interesting statistics:

Over the past week or so I've played 82 games. Of those games only 12 have resulted in one team's deaths being within 50% of the other teams' deaths, i.e. 70 out of 82 games have been near-total wipes in one direction or the other.

85% severely unbalanced game results cannot mean that the match maker is working as intended. Granted, 82 games is only a tiny sample size but I think it's enough to reveal that clearly there's a problem. This data doesn't say what that problem might be, but very obviously there is one.

Actually it can. The Match makers takes you Elo and averages it in with 11 others. Now if you are that teams High Elo and you die. What does that do to the chances of your team winning? Also Elo cannot calculate for attrition. Everything MAY start out balanced, but lose one team mate and the Elo balance is shot!

DeathlyEyes, on 07 April 2015 - 04:12 PM, said:

Reduce the number of players to make individual performance more important. This should make more balanced matchmaking.

And greater cries cause I can't lose, I have Skillz!

[Sir Wulfrick]

Joseph Mallan, on 09 April 2015 - 08:29 AM, said:

Actually it can. The Match makers takes you Elo and averages it in with 11 others. Now if you are that teams High Elo and you die. What does that do to the chances of your team winning? Also Elo cannot calculate for attrition. Everything MAY start out balanced, but lose one team mate and the Elo balance is shot!

True, I hadn't considered that. I think it would be interesting to see an in-depth analysis of how exactly the match maker behaves:

Example 1: MM calculates a mean or median ELO for an available group, sees that another available group isn't far off and grabs some extra players with outlying ELO numbers to even up the mean or median ELO numbers for the entire group and then commits them all to a game.

Example 2: MM finds two groups of 12 using mean or median ELO of iterated sets of 12 players, and when it comes up with the best possible match, commits all 24 of those players to a game.

I could probably think of a few other variations but those were the first that came to mind. Given the way in which the teams are populated, especially for CW matches, I would hazard a guess that the MM is using something akin to the 1^st example, which might also go some way towards explaining the perceived phenomenon of ELO hell - something that I'm increasingly inclined to believe is unfortunately real.