Jump to content


when are you starting to include the PR in your MM


  • Please log in to reply
37 replies to this topic

yatano #1 Posted 24 February 2018 - 08:23 AM

    Lance-corporal

  • Player
  • 18388 battles
  • 55
  • [D-S-S] D-S-S
  • Member since:
    09-21-2013

I don't even talk about xvm, screw that for the vanilla people, but when are you starting to include your very own PR into the mm? Similar to the ranked battles matching battles by rank, you should be doing so with your PR so that the matches are geting more balanced over a long run. I know this won't be possible in every match but why do there have to be 5 super OP player in the enemy team when there are only noobs in the other team?

what is the difficulty in balancing it more?

 

or has this already been chewed to death?



Babbet_1 #2 Posted 24 February 2018 - 08:30 AM

    Warrant Officer

  • Player
  • 10563 battles
  • 719
  • [WJDE] WJDE
  • Member since:
    12-13-2015
Yes! :)

Squirting_Elephant #3 Posted 24 February 2018 - 08:32 AM

    Sergeant

  • Player
  • 9773 battles
  • 219
  • [VIVU5] VIVU5
  • Member since:
    06-10-2011
It has been chewed to death + I also asked this yesterday :P. But its a bit more complicated. Otherwise I would tank my rating in some tier 1 tank all day long and then play tier 10 matches.

HugSeal #4 Posted 24 February 2018 - 08:43 AM

    Captain

  • Player
  • 22717 battles
  • 2,010
  • [SWEC] SWEC
  • Member since:
    05-10-2012
It has. There are several competitive game modes. Random isn't one of them. 

yatano #5 Posted 24 February 2018 - 09:35 AM

    Lance-corporal

  • Player
  • 18388 battles
  • 55
  • [D-S-S] D-S-S
  • Member since:
    09-21-2013
the lower tier grind wouldn't really matter cause u'd still meet like lvld players, and besides that's happening already and you would see the played tier lvl which isn't really making any sense too...and the competitive modes don't really affect the random you still want balanced matches, competitive has nothing to do with balance so that point is a bit invalid...

RamRaid90 #6 Posted 24 February 2018 - 09:41 AM

    Lieutenant General

  • Player
  • 20586 battles
  • 6,285
  • [T-D-U] T-D-U
  • Member since:
    12-14-2014

Clueless people want the teams balancing by PR (which is an absoutely meaningless metric)

 



Long_Range_Sniper #7 Posted 24 February 2018 - 09:50 AM

    General

  • Player
  • 30656 battles
  • 8,276
  • [-MM] -MM
  • Member since:
    04-04-2011

View Postyatano, on 24 February 2018 - 07:23 AM, said:

or has this already been chewed to death?

 

It has.

 

Personal Rating includes your winrate as part of the metric.

 

So if you balance by PR, then what happens to everyone's winrates?



Laur_Balaur_XD #8 Posted 24 February 2018 - 09:54 AM

    Warrant Officer

  • Player
  • 35393 battles
  • 640
  • Member since:
    04-20-2011

View Postyatano, on 24 February 2018 - 07:23 AM, said:

I don't even talk about xvm, screw that for the vanilla people, but when are you starting to include your very own PR into the mm? Similar to the ranked battles matching battles by rank, you should be doing so with your PR so that the matches are geting more balanced over a long run. I know this won't be possible in every match but why do there have to be 5 super OP player in the enemy team when there are only noobs in the other team?

what is the difficulty in balancing it more?

 

or has this already been chewed to death?

 

You want skill based MM? Go play ranked. Or go play SH and/or Advances.

Skill based MM in random would be the end of this game.



spamhamstar #9 Posted 24 February 2018 - 09:56 AM

    Second Lieutenant

  • Player
  • 64310 battles
  • 1,437
  • [LLAY] LLAY
  • Member since:
    12-02-2012

View Postyatano, on 24 February 2018 - 08:35 AM, said:

the lower tier grind wouldn't really matter cause u'd still meet like lvld players, and besides that's happening already and you would see the played tier lvl which isn't really making any sense too...and the competitive modes don't really affect the random you still want balanced matches, competitive has nothing to do with balance so that point is a bit invalid...

 

So you could just run a bot during the day to ruin your stats & then get easy games all night when you're actually playing?

 

The main problem tho is that it's just not fair on the better players.  Why should they get punished with progressively worse mm just for improving?  Unicums would be pretty much guaranteed teams full of tomatoes every game due to the larger number of players of limited ability.

 

You would also find that everyone would gravitate towards a 50% win rate, as poorer players were carried more & better players had less games they could dominate.  There's also questions about how good a metric PR is for measuring ability as LRS just said.

 

View Postyatano, on 24 February 2018 - 07:23 AM, said:

I don't even talk about xvm, screw that for the vanilla people

 

Why would it bother vanilla people more than anyone else, they still have wn8 & eff ratings which could just as easily be used for skill mm if that was a good thing.  They can also check everyone else's wn8's if they want.



vasilinhorulezz #10 Posted 24 February 2018 - 11:21 AM

    Second Lieutenant

  • Player
  • 22780 battles
  • 1,097
  • Member since:
    09-26-2014
No.

_EXODUZ_ #11 Posted 24 February 2018 - 11:25 AM

    Lieutenant

  • Player
  • 34635 battles
  • 1,936
  • [RGT] RGT
  • Member since:
    11-05-2014
I've seen full red vs. red teams many times. It doesn't stop one sided games. Players that have balls and aggression win games.

piratha #12 Posted 24 February 2018 - 11:45 AM

    Corporal

  • Player
  • 8855 battles
  • 131
  • [2-KP] 2-KP
  • Member since:
    10-07-2017

View PostLong_Range_Sniper, on 24 February 2018 - 09:50 AM, said:

..cut..

 

So if you balance by PR, then what happens to everyone's winrates?

 

Just divide them in classes, like for the tank composition: 70% below average players, 20% average and the rest uber-players. Just mix them with different battle configurations, the same as mono-tiers also mono-skill classes, or +/-2 Tiers --> also +/- 25% skillset distribution (or any kind of balanced distribution of skills among teams).

 

Same-skill battle might tend to level the winrate for everybody, but it doesn't have to be always. The question is: will it blend? :P


Edited by piratha, 24 February 2018 - 11:46 AM.


Long_Range_Sniper #13 Posted 24 February 2018 - 12:13 PM

    General

  • Player
  • 30656 battles
  • 8,276
  • [-MM] -MM
  • Member since:
    04-04-2011

View Postpiratha, on 24 February 2018 - 10:45 AM, said:

Same-skill battle might tend to level the winrate for everybody, but it doesn't have to be always. The question is: will it blend? :P

 

Whichever way you do it, balancing teams either within MM, or after MM has selected the team and then swap players between the teams will tend to drive global winrates towards 50%. As long as winrate is in the selection metric somewhere then no matter which way you do it you will tend towards 50% for everyone. All that you can influence is the glidepath on the gradient and time to get there.

 

The other aspect to consider is that XP/Credits are weighted towards winning being the prime objective. The game is based on progression through tech trees with upgrading your tanks and crews an objective to aspire to in order to perform better. Even if you don't want to move up a tier, the game allows you to upgrade your tanks within a tier, and also use consumables which costs credits. These aspects can improve your performance.

 

If you balance for skill in randoms then what you're saying to the playerbase is that if you're a bad player you'll earn more credits than you deserve over time, and if you're a good player you'll lose credits over time as your teams will be artificially rigged to give a winchance that's closer to 50% on a per game basis.

 

Balancing per game is not the solution for this type of economy that promotes winning as an aspirational goal and allocates rewards accordingly.

 

Whatever the rights and wrongs of the matchmaker it is balanced over time in that everyone has the same chance of getting good and bad teams (unless you're convinced Wargaming pick on you personally). 

 

This means that the rewards you achieve in terms of XP and credits are in direct proportion to your results and are attributable to you and not the fact you were carried to a win by better players. 

 

The solution in a random MM to get better results is to get better at identifying your strengths and weaknesses in the game. Charging through the tiers and expecting MM to compensate for your lack of knowledge and ability should not be the fault of the game. Every player will always be able to learn something about the game at each tier and with each vehicle class. If more players stayed at the tiers they can perform reasonably well in before they stretched themselves then I think we'd get less threads asking for an MM based on skill. 



piratha #14 Posted 25 February 2018 - 02:08 AM

    Corporal

  • Player
  • 8855 battles
  • 131
  • [2-KP] 2-KP
  • Member since:
    10-07-2017

Same tier battle was not my final suggestion, it was an example/parallel with mono-tier battles. What happens if among 15 players on each side, you can have an equal weighted distribution of skills among tiers? I've seen (battles) where a Lowe on one side had a <40% WR, the one on the other side had blue digits; and the example is not limited to that.

 

Wouldn't be better if both 15-players could count on (+/-) same amount of tank/tier/types and a balance among the skillset? This will allow all kind of players to mix in the same battle, but without having the good player(s) in a M5 Stuart on one side plus wasting totally the higher tier slot with an unskilled one. I'm just talking about balancing the skills among team, not just the tank tier/type.

 

But I think the MM algo will never converge.. it could be a start if it could match starting from higher tiers going down with less and less stringent range-balance (logaritmic scale). :great:


Edited by piratha, 25 February 2018 - 02:10 AM.


_T_1_T_4_N_0_ #15 Posted 25 February 2018 - 02:56 AM

    Second Lieutenant

  • Player
  • 29068 battles
  • 1,114
  • [SQCI] SQCI
  • Member since:
    02-17-2015

View PostSquirting_Elephant, on 24 February 2018 - 08:32 AM, said:

It has been chewed to death + I also asked this yesterday :P. But its a bit more complicated. Otherwise I would tank my rating in some tier 1 tank all day long and then play tier 10 matches.

 

Its not complicated at all the hall of fame has individual tank ratings based on last x number of battles.
If they bring that in I may even start playing again but the ROFLstomp meta that is now aint for me.

All it needs is an optional filter to state you will wait to join an adaptive bandpass filter based on tank rating which centers on Fifo and widens on queue time.
Its just a filter over standard MM and its really needed if you like more balanced competitive games. 
 

Balc0ra #16 Posted 25 February 2018 - 03:10 AM

    Field Marshal

  • Player
  • 64429 battles
  • 15,448
  • [WALL] WALL
  • Member since:
    07-10-2012
Skilled based MM in randoms already killed one game. No need to do it to this too. Then again that's what ranked is for. As WG said, randoms will remain random. As if you pre set your team based on skill... it's no longer random.

Edited by Balc0ra, 25 February 2018 - 03:10 AM.


brumbarr #17 Posted 25 February 2018 - 06:35 AM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012
Have you played ranked? There are way, and i do mean way more roflstomps there than in randoms. And ranked has skillbased MM.

RamRaid90 #18 Posted 25 February 2018 - 07:54 AM

    Lieutenant General

  • Player
  • 20586 battles
  • 6,285
  • [T-D-U] T-D-U
  • Member since:
    12-14-2014

View Postbrumbarr, on 25 February 2018 - 05:35 AM, said:

Have you played ranked? There are way, and i do mean way more roflstomps there than in randoms. And ranked has skillbased MM.

 

Every game is a damn roflstomp in ranked.

 

You either end up with 6 Object 907s...or you lose...



Jigabachi #19 Posted 25 February 2018 - 08:15 AM

    Field Marshal

  • Player
  • 17858 battles
  • 18,542
  • [T-D-U] T-D-U
  • Member since:
    08-12-2011

View Postyatano, on 24 February 2018 - 08:23 AM, said:

or has this already been chewed to death?

Yes multiple times. But the problem always is that half the people chewing on it have no idea what skillbased MM actually means.



TankkiPoju #20 Posted 25 February 2018 - 08:33 AM

    Lieutenant General

  • Player
  • 20014 battles
  • 6,192
  • Member since:
    05-20-2011

If you want ranked to be "skill based", the ranked MM should put same tanks on both sides, not just people with same ranks.

 

I don't know if that would make very interesting games, but it would be more balanced.






1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users