Jump to content

Public Test - For Science! - Matchmaker


53 replies to this topic

#21 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 24 June 2014 - 05:14 PM

Should be much longer waits before the release valves kick in. Solo matches.

Edited by Bilbo, 24 June 2014 - 05:55 PM.


#22 Veritae

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 269 posts

Posted 24 June 2014 - 05:22 PM

I would love to test this, but the keybind bugs are preventing me from playing as usual on live. I finally figured out that a Triple-Click in some cases will let you modify a key bind. But things that I have bound on live, such as my joystick throttle, is unbindable on PTS. If I cannot replicate my usual playstyle, I can't even begin to determine whether or not the new systems help or hinder.

Le Sigh.

Back to Live Server I go,

#23 KriZ

    Member

  • Pip
  • Elite Founder
  • Elite Founder
  • 12 posts
  • LocationGermany

Posted 24 June 2014 - 05:23 PM

Done a few matches in a heavy. searching took more than 4min almost all the time. as im writing this i am actually searching for a match for about +9min now. matches done at about 6pm from the EU.

Edited by KriZ, 24 June 2014 - 05:24 PM.


#24 BlazeOn

    Member

  • PipPip
  • The Undertaker
  • The Undertaker
  • 30 posts
  • LocationCanada

Posted 24 June 2014 - 05:59 PM

Dropped solo, here is the match. http://i.imgur.com/aEH0vHS.jpg 8 assaults

#25 MonkeyCheese

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,045 posts
  • LocationBrisbane Australia

Posted 24 June 2014 - 06:03 PM

Searching times were fine solo and in a 2man, and I know this is a test server population and all but I just solo dropped into a match where the enemy team had 8 assaults

Posted Image

I hope you guys are getting valuable data to tweak with because at least in that match the rule of 3/ 3x4 thing failed completely


Edit just got another in my next match i am currently in
Posted Image
Dont even know the other team yet...

Edit- other team had 5 atls

Edited by MonkeyCheese, 24 June 2014 - 06:13 PM.


#26 BlazeOn

    Member

  • PipPip
  • The Undertaker
  • The Undertaker
  • 30 posts
  • LocationCanada

Posted 24 June 2014 - 06:09 PM

View PostMonkeyCheese, on 24 June 2014 - 06:03 PM, said:

Searching times were fine solo and in a 2man, and I know this is a test server population and all but I just solo dropped into a match where the enemy team had 8 assaults

Posted Image

I hope you guys are getting valuable data to tweak with because at least in that match the rule of 3/ 3x4 thing failed completely


Edit just got another in my next match i am currently in
Posted Image
Dont even know the other team yet...

We did not fail however :)

#27 Abivard

    Member

  • PipPipPipPipPipPipPipPip
  • Shredder
  • 1,935 posts
  • LocationFree Rasalhague Republic

Posted 24 June 2014 - 06:23 PM

it does seem to be placing the same number of assaults on each team.

#28 AnimeFreak40K

    Member

  • PipPipPipPipPipPip
  • The Defiant
  • The Defiant
  • 455 posts
  • Twitter: Link
  • LocationSomewhere between the State of Confusion and the State of Insanity.

Posted 24 June 2014 - 06:47 PM

View PostAbivard, on 24 June 2014 - 06:23 PM, said:

it does seem to be placing the same number of assaults on each team.

Thinking about it, if the 3/3/3/3 doesn't work, I see nothing wrong with making sure that the number of a given weight class is equal for both teams

#29 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 24 June 2014 - 06:55 PM

At the beginning and end I thought about switching weight classes.

#30 MonkeyCheese

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,045 posts
  • LocationBrisbane Australia

Posted 24 June 2014 - 06:59 PM

View PostAnimeFreak40K, on 24 June 2014 - 06:47 PM, said:

Thinking about it, if the 3/3/3/3 doesn't work, I see nothing wrong with making sure that the number of a given weight class is equal for both teams

We had that way back in closed beta and it was glorious, If I was the only light on a team I would 100% know that there was only one light on the enemy team.

#31 Sylvara-Arc

    Rookie

  • Urban Commando
  • Urban Commando
  • 7 posts

Posted 24 June 2014 - 07:00 PM

In both the Solo(2-4) and Group(2-4) queues, wait times were fairly quick generally around a minute or so. Though none of the matches, not a single one had the full 3x4. Though in near all cases barring a few select rounds each weight class was balanced across both teams.

#32 AnimeFreak40K

    Member

  • PipPipPipPipPipPip
  • The Defiant
  • The Defiant
  • 455 posts
  • Twitter: Link
  • LocationSomewhere between the State of Confusion and the State of Insanity.

Posted 24 June 2014 - 07:04 PM

View PostSylvar, on 24 June 2014 - 07:00 PM, said:

In both the Solo(2-4) and Group(2-4) queues, wait times were fairly quick generally around a minute or so. Though none of the matches, not a single one had the full 3x4. Though in near all cases barring a few select rounds each weight class was balanced across both teams.

That was something I noticed too. Honestly, I think I like this idea much better than a strict implementation of the 3/3/3/3 thing. However, given the way other parts of this test has gone (Falling/Collision damage), I don't see Lights seeing very much play time after this patch goes live.

#33 Acrius

    Member

  • PipPipPip
  • Mercenary
  • 77 posts

Posted 24 June 2014 - 07:05 PM

Dropping solo the wait times were certanily longer then the live servers, but only by a couple of minutes for me. In 10 or so games, picking lights (low pop) did not appear to change the wait time much over heavy (the highest pop); however, it did appear to follow the rule of 3s. Heavy drops never did.

#34 vortmax

    Member

  • PipPipPipPipPip
  • 100 posts
  • Twitter: Link
  • LocationNorman, OK

Posted 24 June 2014 - 07:05 PM

Solo queue dropped in 1-2 minutes. Even sync-dropped with another.

Group of 2: Light/Assault cancelled after 8 minutes. Light/Light cancelled after 9 minutes.

#35 Strongpaw

    Member

  • PipPipPipPipPip
  • Fire
  • Fire
  • 100 posts
  • LocationSouth of Montreal

Posted 24 June 2014 - 07:18 PM

Ok 3F 3A 3I 3L is and always will be a bad concept that does not, can not and will never work! Matchmaker should always deploy on balancing Total tonnage as close as it can get +-50 tons difference between both sides.

(If 24 people are playing in atlases then balance it out at +-0 tons with 12 Atlases on each side.)

Here's why 3/3/3/3 does not work.

Team A;

3 Locust's = 60 tons
3 Cicada's = 120 tons
3 Dragon's = 180 tons
3 Victor's = 240 tons

So team A deploys at 600 tons

Team B;

3 Jenner's = 105 tons
3 Griffin's = 165 tons
3 Orion's = 225 tons
3 Atlas's = 300 tons

So team B deploys at 795 tons

3/3/3/3 fair? Team B is deploying with 195 tons more weaponry that Team A. There is no Balance in this system. There is no Fairness in this system.

KISS Solution? Balance by Weight Deployment.

Strongpaw

#36 Jman5

    Member

  • PipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 4,914 posts

Posted 24 June 2014 - 07:19 PM

The 5-10man search queues seemed comparable to live server's 12 man queue which meant they were sometimes long, but tolerable. In my mind this means once you get this to the live population I imagine search times will fall dramatically.

Elo didn't feel like it was matching very well, but I hope that is simply a low population PTR issue.

Overall, this was a really positive addition to the game. As people logged onto our teamspeak we were able to seamlessly bring them into our group which was really great. We all had a great time and there was no juggling people into and out of small premades like we normally have to do.

What I am most excited about is being able to keep playing when your 12-man lose a couple guys without having to stop everything to scrounge for players. (or worse break up completely)

Edited by Jman5, 24 June 2014 - 07:22 PM.


#37 Deathlike

    Member

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

Posted 24 June 2014 - 07:31 PM

For the purpose of science, please read this:
http://mwomercs.com/...ost__p__3503937

The new MM was disabled during the test, so I'm trying to ascertain when that occurred. Please consider the time of when such stuff occurred because it was disabled (Karl's post is 2.5 hours past the start of the test).

#38 DeathlyEyes

    Member

  • PipPipPipPipPipPipPip
  • The Messenger
  • 940 posts
  • Google+: Link
  • Facebook: Link
  • Twitter: Link
  • Twitch: Link
  • LocationMetaphorical Island somewhere in the Pacific

Posted 24 June 2014 - 07:31 PM

3/3/3/3 won't work. Do this instead.

View PostSLDF DeathlyEyes, on 20 June 2014 - 03:08 PM, said:

Right now the current way the ELO system works is totally unacceptable. It tries to balance entire team ELOs often times using high ELO players to balance out low ELO players to make matches. This doesn't work. To make things worse the matchmaker doesn't take into account the tonnage the high ELO players are in. This means that someone with a high ELO could be expected to make up for low ELO players in high tonnage against a team of heavy tonnage above average ELO players. Personal experience shows that this leads to matches like this:

Posted Image


Not very fun. There is no way Apostal and I could make up for all the low ELO players on our team, especially in the light mechs we were in. We were the only 2 grouped and we ended up in Alpha lance against 2 known high ELO groups. This leads me to my suggestions.

Instead of using tonnage balancing the game should use a battle value system. This could take into account a mech's tonnage, armor, engine size, weapon types, weapon combinations and equipment. This would allow for some sorting based on if a mech is "meta" or not. Automatically battle value could be raised based off if a mech has a combination of PPC+AC+Jump Jets. For example a meta multiplier of .25 could be applied if a mech is a heavy/assault, has jump jets and has 2 PPCs. So the battle value could be calculated by doing (mech BV)+((meta multiplier)(mech BV)). This would then start to seperate meta mechs from the rest.

The next step to solving the problem is to calculate a player's battle effectiveness. Just because a player has a high ELO in a mech doesn't mean he/she is going to be super effective on the battlefield in every mech. This would also open up more play options for players to use less optimized builds and not have to worry about someone not knowing what he/she is doing and dropping in a high BV mech, completely skewing the matchmaking. A players combat effectiveness could be calculated by a calculation such as (Mech BV)(.5)(ELO). This would give out a number that could more closely equate to how effective this player is on the battlefield allowing him/her to leave meta without fear that he/she's high ELO would be needing to making up for low ELO players playing in high BV mechs. This may become difficult at non peak hours so things might not be entirely perfect and need to be loosened up. This should help with peak hours and prevent matches from getting to ridiculous.

Finally edge case ELOs should be reset every month or with any major balance changes. The top 20% of all players should have their ELOs automatically reset to the 80th percentile to let them have a chance to be reorganized after any substantial meta changes. The same should be done with the bottom 20% as it will give them a chance to crawl out of low ELO. Their substandard builds could also become viable from the change and this would further mitigate the skewing of balance changes on the matchmaker.

The next change is how the matchmaker assembles teams. The matchmaker should try to match Combat Values of each party entering into the match. Each group should have a similar combat effectiveness level to one another so that one group isn't making up for a huge difference between another group.

Now these numbers are just examples of what could work. I do not have access to all the lovely statistics Paul has access to but I do have experience with the issues at hand. These changes could drastically improve the game play experience for everyone.


#39 Strongpaw

    Member

  • PipPipPipPipPip
  • Fire
  • Fire
  • 100 posts
  • LocationSouth of Montreal

Posted 24 June 2014 - 07:37 PM

Loooonnggg! Sloooooooowww! Slow as in minutes slow. I solo drop 99.9% of the time. Even taking a light mech. Its not a good thing when I can pick up a book a read a chapter in it before I get a connecting to screen match.

Strongpaw

#40 Agent 0 Fortune

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 3,403 posts

Posted 24 June 2014 - 07:51 PM

I only played a few matches at the tail end of the public test.
I dropped solo every time, and changed my mech weight class most drops.

Drop Times for queue:
Solo drops in a highly populated weight class took around 1 min to find a match. And just my luck that every time I changed weight class between matches, I was unfortunately enough to coincidently select the overpopulated class (yes even mediums got up to 43% when I dropped there). This could almost be mitigated with a 4 mech quick-drop deck (lostech maybe).

4x3:
I believe 1 match had any semblance of 4x3 (oddly enough it was one of the longer wait periods, so probably coincidental). One particularly bad match saw 8 Assaults on our team.

Solo Queue:
I do not believe any teams snuck into the solo queue, in fact on the few games where I saw the same faces in back-to-back matches, the those players switched team (I almost suspect that the matchmaker went out of its way to make sure if two players were on opposite teams last game, they were on the same team next game, as this occurred a couple times).

ELO:
Didn't play enough games, but I believe I had close to an even record.





7 user(s) are reading this topic

0 members, 7 guests, 0 anonymous users