Jump to content


how come you don't take skill/win rate into account in matchmaking


  • Please log in to reply
1958 replies to this topic

BillT #21 Posted Jan 23 2018 - 22:54

    Major

  • Beta Testers
  • 27607 battles
  • 5,104
  • [F-3] F-3
  • Member since:
    08-13-2010

View Postcaspertoo, on Jan 23 2018 - 16:29, said:

The new matchmaking has some serious flaws.  One of which is it's inability to take skill level into account when making teams.  Wargaming has some programmers that are capable of tying their own shoes, I hope, so why are there so many games that are so heavily uneven?  When you have 4 blue players, 2 purple players, and a ton of green players vs 3 green players and the red yellow and below, it is indicative of a serious problem with this game and it makes the game seriously not fun at all.  

 

So instead, do you really want one match with all purples and blues, another with greens and yellows, and a third with all oranges and reds?  Bear in mind that half the players in each match lose.  So those purple players see their win rates pushed down toward 50%, and the reds see theirs pushed up toward 50%.  Your 59% win rate would probably end up around 53% or 54% - still above average, but not so far above average.  Most top players don't like the sound of this.

 

Edit: I should add that this happens no matter how you balance the two teams.  If each team gets 1 purple, 2 blues, 3 greens, 3 yellows, 3 oranges, and 3 reds, half of the purple players in the match still lose, and half of the reds still win.


Edited by BillT, Jan 23 2018 - 23:00.


HisNameWasSethRich #22 Posted Jan 23 2018 - 22:54

    First lieutenant

  • Players
  • 14549 battles
  • 556
  • Member since:
    08-09-2012

View Post_Gungrave_, on Jan 23 2018 - 16:53, said:

 

Fixed

 

This guy gets it.  ^^^

DrOoLeR_In_Da_HoUsE #23 Posted Jan 23 2018 - 22:55

    Major

  • Players
  • 41380 battles
  • 3,595
  • [SUX] SUX
  • Member since:
    10-09-2011

View PostHisNameWasSethRich, on Jan 23 2018 - 15:52, said:

 

(((Personally, I have not played in ranked))) 

Again, as my good friend Dominator_98 mentioned, They turned the mode into just a grind fest over it being a skill based ranking system.  

The issue with basing any type of SBMM off PR is that PR is an easily padded stat. 

 

no really... that's why u don't do SBMM... I dare YOU to show me somebody with a padded personal rating that doesn't represent their skill.... But I sure as heck can show a padded WN8 :P

Hurk #24 Posted Jan 23 2018 - 23:01

    Major

  • Players
  • 55301 battles
  • 17,375
  • [KGR] KGR
  • Member since:
    09-30-2012

there are 5 types of skill matching:

 

1. total team (team A and B should have 15000 "points" )

2. total team weighted (Team A and B should have 15000 "points", but a low tier tank or SPG is worth less points, while a meta medium high tier tank is worth more, etc)

3. individual weight, with gains and losses based on competition and action. (ELO ratings, where you gain more points for taking out more skilled players, and lose more points when you lose to weaker players, etc)

4. individual weight with team bias. (your team affects the ELO outcome, if purples carry you, you gain less points, if you die to a carried team, you lose less points, etc)

5. laddered play. (players who win X percent of the time on this ladder, move to the next ladder, and some form of ELO is used to determine when you drop or go up)

 

the issue with every single one of these is that they make random games no longer random. and they make queue times sky rocket since we dont have a large pile of players waiting to play at every tier on the NA cluster.

 

so.... if we reach 100,000 people in the queue, then yes, SBMM can be applied,  since you have enough people to make it "fair". otherwise, the NA cluster can only tolerate type 1/2 since its still random matching for the most part, with only gains/losses being adjusted after the fact. total impact to matching would be minimal, since you would not try very hard in the first place.  (since the handicap after the match to scoring handles the discrepancies)

 

 

that said, i dont believe that every match being a real challenge will make the game better, in fact, it would flatten the tiering system to the point where it would be irrelevant, since the competition in tier 5 tanks would become just as competitive as tier 8+, since if you queued, the matcher would try to make it so. 

I believe if the game had launched that way, and the idea or crawling up tiers was presented as instead a widening of the tank pool to play from, we would not have as many skill failures as we do now. 

 

but as everyone else will point out, the win rates would also flatten out considerably. as people face similar ELO ratings more often, their win loss drops to 50%.  for many people the pull of the game is being able to seal club to a degree, or on occasion. (I myself still really like playing low tier with all its chaos, but yes, skill and crews and equipment makes it a lot easier)


Edited by Hurk, Jan 23 2018 - 23:03.


Slim_Shadee #25 Posted Jan 23 2018 - 23:05

    First lieutenant

  • Players
  • 6464 battles
  • 764
  • [USAA] USAA
  • Member since:
    03-04-2011

View PostholdmecloserTonyDanza, on Jan 23 2018 - 21:43, said:

 

based on what? because people say so?  how the [edited]could it be any worse than the [edited]that goes on in pubs everyday?  not saying it's the answer, but I love all these people that scream from the rooftops "IT CAN"T WORK" with absolutely zero proof of what could happen because it's never been tried...same as the the "just because they have the patent doesn't mean they use it" crowd

 

it's no wonder this company is worth a billion dollars, this player base would literally swallow anything they get fed

 

Has been tried on here in one form called Ranked and in that failed other tank game. Doesn't work. It solves nothing except increase wait times for the right match. Right tank, right player, right map win rate, time of day, the variables become to lengthy to make a decent time frame for wait time. Add to the the variables that have to be considered to "eliminate" the complaints that will surely arise its Nirvana that solves nothing and adds issues.

 

View PostholdmecloserTonyDanza, on Jan 23 2018 - 21:48, said:

 

ranked?  that's not sbmm..what?  I've seen 47% players who played them enough to advance ranks

 

This is not a reason for adding a SBMM in the game. The fact it simply wont change anything based on the population and current average win rate on the server. You talk about 47% win rate the average is 48% on the server which means the bulk of the MM you want will not change. Take some time and actually look at a random number of matches 300 or so your in and you see that there is no magic in having two or three high win rate players on your team. What does effect MM outcome is the matches you have clan platoons working thru TS coordinating their fire and map placement. The fallacy of SBMM is without this the random high win players magically improve the match. Yet constantly I hear on here I cant carry 10 tomatoes. Well how many "better" win rate players do you think skill MM is going to add to your team? Second those players with a high win rate have achieved it in the current system. Some legit some by working the system. How do you ferret out those that didnt get legit numbers? Then you open that can of worms. Lets go SBMM and increase the wait time of the better players so they become frustrated and get on here complaining that if they were red they would get right into a match instead SBMM has them waiting minutes for that match. Again its population issue and various other factors that will make it so frustrating for better win rate players they quit.

 

Post 44 million on this fantasy idea.



R_Razor #26 Posted Jan 23 2018 - 23:07

    Captain

  • Players
  • 8095 battles
  • 1,757
  • [_DDM] _DDM
  • Member since:
    08-20-2013

View PostHisNameWasSethRich, on Jan 23 2018 - 16:31, said:

Skill based MM does not work. This has been argued already.

 

Mechwarrior Online proves you wrong but don't let that stop the standard responses. 

R_Razor #27 Posted Jan 23 2018 - 23:08

    Captain

  • Players
  • 8095 battles
  • 1,757
  • [_DDM] _DDM
  • Member since:
    08-20-2013

View PostHellsfog, on Jan 23 2018 - 16:37, said:

Do a forum search for the other 4,593,584,701,584, thread on the topic. 

 

Why? The Anti Arty crowd doesn't, what makes this topic any different?

 



DrOoLeR_In_Da_HoUsE #28 Posted Jan 23 2018 - 23:09

    Major

  • Players
  • 41380 battles
  • 3,595
  • [SUX] SUX
  • Member since:
    10-09-2011
its like talking to a wall in the forums.. kinda like pub play

holdmecloserTonyDanza #29 Posted Jan 23 2018 - 23:11

    Major

  • Players
  • 67432 battles
  • 6,846
  • [HSOLO] HSOLO
  • Member since:
    08-21-2011
RANKED BATTLES IS NOT SBMM

HisNameWasSethRich #30 Posted Jan 23 2018 - 23:11

    First lieutenant

  • Players
  • 14549 battles
  • 556
  • Member since:
    08-09-2012

View PostSir_Tomato_DroolsAlot, on Jan 23 2018 - 16:55, said:

 

no really... that's why u don't do SBMM... I dare YOU to show me somebody with a padded personal rating that doesn't represent their skill.... But I sure as heck can show a padded WN8 :P

 

Again, no one mentioned WN8 here. I agree that WN8 is a padded stat, I'm not gonna argue that. Naming someone with a padded PR would be naming and shaming, but I can give an example. Someone with, of IDK, 30K games, would have, on average a PR about 1500 points higher than a player with very similar resents who has just 15K games. 

HisNameWasSethRich #31 Posted Jan 23 2018 - 23:12

    First lieutenant

  • Players
  • 14549 battles
  • 556
  • Member since:
    08-09-2012

View PostholdmecloserTonyDanza, on Jan 23 2018 - 17:11, said:

RANKED BATTLES IS NOT SBMM

 

It was meant to be, sadly WG can't do anything right. 

Edit: In reality, you can't prove it's not either, Just because a player with 47% WR is a high rank doesn't mean he's bad? It's his teams that lost him games? isn't that OP's entire point here? 

Edited by HisNameWasSethRich, Jan 23 2018 - 23:13.


holdmecloserTonyDanza #32 Posted Jan 23 2018 - 23:13

    Major

  • Players
  • 67432 battles
  • 6,846
  • [HSOLO] HSOLO
  • Member since:
    08-21-2011

View PostHisNameWasSethRich, on Jan 23 2018 - 16:12, said:

 

It was meant to be, sadly WG can't do anything right. 

 

you will not get an argument from me on the second part of your statement

Slim_Shadee #33 Posted Jan 23 2018 - 23:20

    First lieutenant

  • Players
  • 6464 battles
  • 764
  • [USAA] USAA
  • Member since:
    03-04-2011

View PostR_Razor, on Jan 23 2018 - 22:07, said:

 

Mechwarrior Online proves you wrong but don't let that stop the standard responses. 

 

1, "Matchmaker is completely broken, rendering MWO's quickplay portion almost as un-fun to play as FaP: Either you dish out or you receive a curb stomp. And while in Fap, some groupe might enjoy this, I do not."

2. It seems pretty clear to me they have reverted the changes they made in early January when they tightened it up.

3. Don't even try. Nothing has been fixed.

I decided to try to play several matches, after some fixes for matchmaker were announced recently. What I found out:
1) Terrible matchmaking - check
2) "Bring terrible map" system - check.

4.

Stopped playing MWO in December, because MM doesnt work and put Tier4 Players into Games with Tier1 Players.
Not enaugh that PSR is broken, but now it does not work even more as newbprotection.
PGI told us, that they fixed it, but when i talk to friends they tell me that the Games are still horrible.

If u cant fix it, get rid of it. thxbb

 

Random accolades from the SBMM in Mech Warrior which is down to 5000 average population of players due to issues that game has with its SBMM an it continues to decline.

Its a rousing YES lets add another nail to the WoT coffin.



Buttknuckle #34 Posted Jan 23 2018 - 23:23

    Major

  • Players
  • 50063 battles
  • 2,904
  • [-RECK] -RECK
  • Member since:
    03-19-2013

View PostHisNameWasSethRich, on Jan 23 2018 - 22:11, said:

 

Again, no one mentioned WN8 here. I agree that WN8 is a padded stat, I'm not gonna argue that. Naming someone with a padded PR would be naming and shaming, but I can give an example. Someone with, of IDK, 30K games, would have, on average a PR about 1500 points higher than a player with very similar resents who has just 15K games.

 

It seems you believe the games played has more impact on PR than it actually does. The impact of number of games played in the PR calc is essentially flat after about 4K games.

_Gungrave_ #35 Posted Jan 23 2018 - 23:24

    Major

  • Players
  • 43752 battles
  • 16,299
  • [-SRP-] -SRP-
  • Member since:
    12-07-2011

View PostHisNameWasSethRich, on Jan 23 2018 - 22:54, said:

 

This guy gets it.  ^^^

 

Well WG themselves in multiple Q&A sessions when asked about skill based MM have said no.



R_Razor #36 Posted Jan 23 2018 - 23:25

    Captain

  • Players
  • 8095 battles
  • 1,757
  • [_DDM] _DDM
  • Member since:
    08-20-2013

View PostSlim_Shadee, on Jan 23 2018 - 17:20, said:

 

Wall of text

 

Nice wall of text proving nothing. 

HisNameWasSethRich #37 Posted Jan 23 2018 - 23:26

    First lieutenant

  • Players
  • 14549 battles
  • 556
  • Member since:
    08-09-2012

View PostButtknuckle, on Jan 23 2018 - 17:23, said:

 

It seems you believe the games played has more impact on PR than it actually does. The impact of number of games played in the PR calc is essentially flat after about 4K games.

 

no... But. Ok. 

Something i didn't even mention was Team Battles. Just about 2 weeks ago I did about 6 games in Team Battles, won all of them and my PR changed by about 30 points. 

Slim_Shadee #38 Posted Jan 23 2018 - 23:31

    First lieutenant

  • Players
  • 6464 battles
  • 764
  • [USAA] USAA
  • Member since:
    03-04-2011

View PostR_Razor, on Jan 23 2018 - 22:25, said:

 

Nice wall of text proving nothing. 

 

Doesn't surprise me about you Razor like your statement proved a damn thing except your propensity for mouthing off with zero facts....but then we dont want another set of mod sending us stop and desist message so I will leave it alone this time.

HAPPYJAX60 #39 Posted Jan 23 2018 - 23:35

    Corporal

  • Players
  • 37412 battles
  • 84
  • [PERCH] PERCH
  • Member since:
    02-19-2012

View PostSir_Tomato_DroolsAlot, on Jan 23 2018 - 21:44, said:

 

you don't base it off win rate.. you base it off PR.. once MM selects the 30 players (no effect on query time), it can then move the players back and forth so that the total PR of both teams falls within +/- 1000 - 2000k PR.. this way each team gets a little sugar and spice, and then you can throw some tomato paste on the side.. :P

its not skill based in the sense that people are thinking where u put all 60% players on both teams, that's just stupid... I am talking balanced, which it totally different.

 

and please don't compare this game to AW.. AW failed for more reason then MM... it was the fact everybody was playing PVE or whatever, and nobody was playing PVP so a lot of people got tired of 10 minute waits for other players.. and bailed.

 

enough with the 6th grand special ed football team vs the New England Patriots

 

but they play the browns

 



da_Rock002 #40 Posted Jan 23 2018 - 23:38

    Major

  • -Players-
  • 8840 battles
  • 3,401
  • Member since:
    11-24-2016

View PostHisNameWasSethRich, on Jan 23 2018 - 16:31, said:

Skill based MM does not work. This has been argued already.

 

 

Ignoring skill makes the present MM worthless.    That hasn't been mentioned much because it'd impact the players who benefit most.

 

 

The huge percentage of roflstomps are the result of naturally occurring skill imbalance when skill is ignored.    They don't have to come up with a perfect SBMM to stop spitting out the volume of WoT/WG signature blowouts.    And they need to stop doing that.   


 

The present template isn't working so good.   Balancing by equipment has never worked so good.   Or maybe they have.   Who would know as long as everyone overlooks the elephant in the room.   That elephant has wrecked every MM system to date.   Because it matters.   Yet WG doesn't even smell it, much less see it.    Or maybe they do. 


 

Those blowouts take very little server time compared to naturally occurring battles.    Ignoring skill turns 40% or so battles into buff teams naturally stomping weak ones.   No matter what template they use or what list of equal tanks is referenced. 


 

They don't need a perfect SB anymore than a perfect equipment list or tier template.   But they do need to consider skill enough that "naturally occurring" happens more often than 60%.






5 user(s) are reading this topic

0 members, 5 guests, 0 anonymous users