Jump to content

We Urgently Need A More Intelligent Matchmaking System


27 replies to this topic

#1 Mcchuggernaut

    Member

  • PipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 838 posts
  • LocationYour core

Posted 13 September 2013 - 04:30 PM

Are there plans for matchmaking improvements any time soon? We need a weight minimum/maximum urgently, or maybe it could just be done by weight class instead. Only so many heavies, lights, mediums, assaults per team? I have seen too many games won or lost by wildly differing team weights. Another thing that needs fixed is veterans dropping with newbies. From the new player side, it can ruin a new player's good time if he/she is constantly getting wasted by better players at the start. From a veteran's perspective, players get frustrated as hell when they lose a match because half of their team is composed of people who don't know what they are doing and all get stomped early, leaving the vet to fight a losing battle against impossible odds. In other games I have played, the matchmaker balanced the teams by putting together a mix of players at different skill levels that tried to balance overall skill between the two sides. It never worked. Good players were rewarded for their skill by constantly having to carry the team of newbies they were saddled with, and newbies were still getting flattened when they met that very skilled player on the opposing team. Halo 2 did it right (I know, that game had nothing to do with Mechwarrior, but bear with me). It had a ranking system, and you ONLY played other people who were within a rank or 3 of yours. The game stayed challenging because you were constantly facing opponents of your skill level, you could rely on your teammates to be nearly as good as you at all times, and good gameplay was rewarded with a rank insignia and the knowledge you were able to earn your rank by getting better playing opponents as tough as you.

Does the community agree with my thoughts on matchmaking? Any suggestions/flaws with my reasoning? Lets hear it!

#2 80sGlamRockSensation David Bowie

    Member

  • PipPipPipPipPipPipPipPipPip
  • Veteran Founder
  • Veteran Founder
  • 3,994 posts
  • LocationThe Island

Posted 13 September 2013 - 04:31 PM

*signed*

#3 Scandinavian Jawbreaker

    Member

  • PipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 1,251 posts
  • Twitch: Link
  • LocationFinland

Posted 13 September 2013 - 04:36 PM

View PostMcchuggernaut, on 13 September 2013 - 04:30 PM, said:

Another thing that needs fixed is veterans dropping with newbies.

Definitely. It is waste of my time and their time when I'm killing trial Cicadas standing still trying to itch me with MGs. I think the best way to avoid this is to seperate the queue; <500 matches as the noob queue and >500 vet queue.

#4 Xandre Blackheart

    Member

  • PipPipPipPipPipPipPip
  • 703 posts
  • LocationIn the "cockpit".

Posted 13 September 2013 - 04:42 PM

Frankly Tonnage was the base mechanic for matching in Table Top.

I see absolutely no reason for the change to some kind of nebulous win/loss ratio as the base mechanic.

You want some fine tuning use your win/loss ratio, but it should always be secondary to tonnage.

#5 The Boz

    Member

  • PipPipPipPipPipPipPipPip
  • 1,317 posts

Posted 13 September 2013 - 04:43 PM

There's a problem.
Elo wasn't meant to work in multiplayer games. ESPECIALLY in 12v12 format. A player has a one twentyfourth the impact on the outcome of a game, but faces the full Elo shift, regardless of personal success in the match.
So... what do you base your matchmaker on?

#6 Deathlike

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 29,240 posts
  • Location#NOToTaterBalance #BadBalanceOverlordIsBad

Posted 13 September 2013 - 04:44 PM

I think many have suggested a "skill based or ladder queue" of some sort... at least in conjunction with a casual/newbie queue. At least that would reduce the issues most of us are citing as issues.

There should be bonuses for the higher level queue... but I'm not entirely sure how that should be applied.

#7 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 September 2013 - 04:51 PM

View PostThe Boz, on 13 September 2013 - 04:43 PM, said:

So... what do you base your matchmaker on?


Exponential moving average of win/loss ratio for last 32 games on a per weight class basis.

#8 Diablobo

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,014 posts
  • LocationOn your six

Posted 13 September 2013 - 04:56 PM

I like the idea of ranks. There can be green, recruit, veteran, and elite. No one plays more than one level away from their rank. Maybe there could be even more ranks, like 6 or even 8 steps.

#9 LethalRose

    Member

  • PipPipPipPipPip
  • 112 posts

Posted 13 September 2013 - 04:59 PM

Id rather have a 2 minute queue and a wide range of player skill on my team than a 10 minute queue with a group more to my skill level.

Its a trade off and I like how fast things are now. I don't want to wait forever when the average game is 7 minutes long.

I feel your pain trust me I guess the best thing to do is play with 3 other people who are close to your skill level.

#10 Mcchuggernaut

    Member

  • PipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 838 posts
  • LocationYour core

Posted 13 September 2013 - 05:03 PM

I don't think you could use someone's win/loss ratio as the matchmaking mechanic. I have more losses than wins, yet I have a positive K/D and am a competent player. Since the matchmaking isn't balanced correctly, K/D and damage done means much more than wins/losses because K/D and damage dealt depends more on your individual skill than whether your team wins does, since as The Boz pointed out, your individual effect on your team is diminished the larger of a group you are put in. Better matchmaking would probably also reduce steamrolls, AND we wouldn't have to play huge 12v12 matches to try and even out the one-sidedness. We could go back to 8v8 where your individual performance statistically matters more and individual good gameplay would more rewarding.

Anyway, these are still just my views and I am open to comments/suggestions/criticism. If I'm wrong, I want to be corrected, and if I have good ideas I hope it benefits the game.

Edited by Mcchuggernaut, 13 September 2013 - 05:10 PM.


#11 Mcchuggernaut

    Member

  • PipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 838 posts
  • LocationYour core

Posted 13 September 2013 - 05:08 PM

View PostLethalRose, on 13 September 2013 - 04:59 PM, said:

Its a trade off and I like how fast things are now. I don't want to wait forever when the average game is 7 minutes long.


It probably wouldn't be that bad if some latitude was given for a rank or three either way being acceptable. There needs to be a middle ground like in all things. Would you be willing to wait 3 minutes for a much more balanced game that still wasn't quite perfect but was as good as reasonably possible? A better balance would probably also mean longer games, too...

#12 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 September 2013 - 05:23 PM

View PostMcchuggernaut, on 13 September 2013 - 05:03 PM, said:

I don't think you could use someone's win/loss ratio as the matchmaking mechanic. I have more losses than wins, yet I have a positive K/D and am a competent player. Since the matchmaking isn't balanced correctly, K/D and damage done means much more than wins/losses because K/D and damage dealt depends more on your individual skill than whether your team wins does, since as The Boz pointed out, your individual effect on your team is diminished the larger of a group you are put in. Better matchmaking would probably also reduce steamrolls, AND we wouldn't have to play huge 12v12 matches to try and even out the one-sidedness. We could go back to 8v8 where your individual performance statistically matters more and individual good gameplay would more rewarding.

Anyway, these are still just my views and I am open to comments/suggestions/criticism. If I'm wrong, I want to be corrected, and if I have good ideas I hope it benefits the game.


Your reasoning is based on a broken matchmaker though. My idea, (EMA on 32 game period win/loss, ranking system) would be responsive. Keep winning games and you keep going up in rank fairly quickly. It reinforces teamwork because doing 750 damage but losing shouldn't be rewarded that much IMO.

Dropping in rank wouldn't be so horrible either because you could gain the rank back fairly quickly. If you got steamrolled 10 times in a row you'd drop so low that if you had any skill you'd probably kill half the enemy team yourself on the next game.

Note that's how I think it should be for PUGs. Premades should have a less responsive matching system (SMA over 64 games for instance) that rewards a long history of wins or punishes a long history of losses by making it harder to gain or lose rank.

Edited by xenoglyph, 13 September 2013 - 05:31 PM.


#13 The Boz

    Member

  • PipPipPipPipPipPipPipPip
  • 1,317 posts

Posted 13 September 2013 - 05:25 PM

View Postxenoglyph, on 13 September 2013 - 04:51 PM, said:


Exponential moving average of win/loss ratio for last 32 games on a per weight class basis.

You're still using a system that places full responsibility for 23 other players on the shoulders of one. And EMAs are NOTORIOUS for their glacial pace.

#14 scJazz

    Member

  • PipPipPipPipPipPipPipPip
  • 1,668 posts
  • LocationNew London, CT

Posted 13 September 2013 - 05:38 PM

What if I told you that the largest weight imbalance I've seen in the last 100 matches was an astonishing 260 tons. Worse, the lighter team had a player Disconnect 70 ton Cataphract. Starting weight penalty 330 tons...

Now what if I told you that the lighter team won... by killing everything.
Posted Image
L2P

kk thx bai o/

#15 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 September 2013 - 05:38 PM

View PostThe Boz, on 13 September 2013 - 05:25 PM, said:

You're still using a system that places full responsibility for 23 other players on the shoulders of one.


It encourages proper teamwork and since it's responsive it's less punitive.

View PostThe Boz, on 13 September 2013 - 05:25 PM, said:

And EMAs are NOTORIOUS for their glacial pace.


What are you talking about? You tailor the EMA equation to be exactly as responsive as you want it to be. An EMA over 256 periods would of course be a lot different than one over 16.

Edited by xenoglyph, 13 September 2013 - 05:39 PM.


#16 The Boz

    Member

  • PipPipPipPipPipPipPipPip
  • 1,317 posts

Posted 13 September 2013 - 05:41 PM

Encourage all you like, you'll never get random pugsleys to reliably participate in a match.
And some will shoot you if you try.

#17 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 September 2013 - 05:48 PM

Well, to be honest I think many different equations should be tried on the test server 24/7. I'm all for testing one that's weighted heavily for damage/kills, another that's weighted heavily for wins, and another that uses a mix. It's not something that can be backtested, it has to be done live.

#18 Hellcat420

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • 1,519 posts

Posted 13 September 2013 - 05:52 PM

unfortunatly its working as intended.

#19 Screech

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • 2,290 posts

Posted 13 September 2013 - 05:56 PM

They were going to impose tonnage limits and get rid of group size limits. They even posted a proposed weight limit table for 2-12 groups. The thought being this system would be easier for the matchmaker as it would then only have to match group size and Elo to make a match.

But that was before 3PV. Since then they have said that 12 mans will be 1PV only so that removes them from being brought back into the general queue. Does that mean they are scrapping the tonnage limits? Are they still removing the group limits on 2-11 and just leave 12mans as is? I don't know, PGI hasn't decided say how the changes are affecting their plans.

#20 Mechsniper

    Member

  • PipPipPipPipPipPip
  • The Formidable
  • The Formidable
  • 457 posts
  • LocationUnited States

Posted 13 September 2013 - 05:59 PM

Sc your lighter lance won largerly due to hit detection errors and the skew of balance leaving light mechs as assault killers. It is sadly easier to kill an atlas in a 3M cicada than in another atlas or Highlander. PGI set things up this way on purpose, and its a fail for game balance. Mediums have just as much trouble with lights and are suicide against assaults, making them currently near useless, and Hard core mode at best. Trashing ELO, the idea that lights should = assaults/Heavies in a firefight and setting up a lobby with in game voip where you set up que's for 4/12 man lances is what is needed. Don't like in game VOIP? Have/use a mute button. TOO EASY!!!! Kill ghost heat, and now you would have a game worth playing. Add: Battle value could be used to balance the lances in the lobby or a weight system.

Edited by Mechsniper, 13 September 2013 - 06:01 PM.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users