Jump to content

Game Balance And Matchmaking


81 replies to this topic

#41 IronWatch

    Member

  • PipPipPip
  • 63 posts
  • LocationCalgary

Posted 26 April 2019 - 05:41 AM

Matchmaker is primarily a placebo effect players use to convince themselves they're losing for a reason. I think the matchmaker on MWO must be good because I'm hovering around a pretty good 50% WR with every mech. This has been a fascinating thread but you all keep posting paragraphs of different facts about how this MM works and I'm peetty certain none of you know for sure.

Edited by IronWatch, 26 April 2019 - 05:42 AM.


#42 McGoat

    Banned -Cheating

  • PipPipPipPipPipPipPip
  • CS 2019 Gold Champ
  • CS 2019 Gold Champ
  • 629 posts

Posted 26 April 2019 - 06:33 AM

One of many suggested methods to improve MM utilizing already logged statistics, Match Score.

They wouldn't even need to really adjust how the MM puts together matches, as the image shows the quality of the games would still improve by placing the most undeserving T1 players in T4 where they belong.

Posted Image

Edited by McGoat, 26 April 2019 - 06:35 AM.


#43 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 26 April 2019 - 07:21 AM

Average match score (MS), which Jarl's uses, doesn't reflect skill. The match score breakdown for actions are arbitrarily decided by PGI instead of being weighed by their contribution to victory so you have things like wasteful damage being rewarded more than clean kills, basically you get the stupid case where between a 1 head shot kill or spraying down a target left and right, you should do the latter because it gives you 10x more MS.

#44 Eatit

    Member

  • PipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 286 posts

Posted 26 April 2019 - 08:43 AM

I like your chart McGoat but the number of tier 1 players on there is pretty small. How will they get a decent wait time with such a low population?

#45 McGoat

    Banned -Cheating

  • PipPipPipPipPipPipPip
  • CS 2019 Gold Champ
  • CS 2019 Gold Champ
  • 629 posts

Posted 26 April 2019 - 09:30 AM

View PostNightbird, on 26 April 2019 - 07:21 AM, said:

Average match score (MS), which Jarl's uses, doesn't reflect skill. The match score breakdown for actions are arbitrarily decided by PGI instead of being weighed by their contribution to victory so you have things like wasteful damage being rewarded more than clean kills, basically you get the stupid case where between a 1 head shot kill or spraying down a target left and right, you should do the latter because it gives you 10x more MS.


It is a far better system than what is currently in place, there is simply no denying that. It may not reflect skill in its entirety, but it does reflect knowledge of the game and its mechanics. A 90%+ player is going to be better than an 80% player in at least the team setting.

In any case, over time the wasteful damage players will gradually fall as those that they're spreading damage against will also slowly be moved down to appropriate levels of play. It is in no way perfect, but perfect doesn't exist.

View PostEatit, on 26 April 2019 - 08:43 AM, said:

I like your chart McGoat but the number of tier 1 players on there is pretty small. How will they get a decent wait time with such a low population?


It's not mine, IIRC Appogge created this version.

As far as T1 players being small, that's ok. The MM functions in such a manner that it slowly opens pools up from T1 v T1, To T1 v T2, and lastly T1 v T3 (interchangeably) until it fills all 12 spots on each side. So while there may be a slightly higher wait for a game, it shouldn't be too much longer and the quality of the match will be higher.

Given that I understand "quality" is an individual metric, but there are some universal truths that genuine higher level players seek in matches.

Edited by McGoat, 26 April 2019 - 09:52 AM.


#46 Racerxintegra2k

    Member

  • PipPipPipPipPipPipPip
  • The God
  • The God
  • 801 posts

Posted 26 April 2019 - 09:48 AM

View PostMcGoat, on 26 April 2019 - 06:33 AM, said:

One of many suggested methods to improve MM utilizing already logged statistics, Match Score.

They wouldn't even need to really adjust how the MM puts together matches, as the image shows the quality of the games would still improve by placing the most undeserving T1 players in T4 where they belong.

Posted Image


A player that avg's 226ms (75th percentile) is far worse than someone averaging 300ms (92th percentile).

I would personally go for something like I've listed below. With a periodic reset. If match time is a problem you open the gates one tier at a time.

anyone below 40th percentile Tier 6
anyone between 50-60th percentile Tier 5
anyone between 60-70th percentile Tier 4
anyone between 70-80th percentile Tier 3
anyone between 85-95th percentile Tier 2
anyone above 95th percentile Tier 1

#47 5th Fedcom Rat

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 893 posts

Posted 26 April 2019 - 10:02 AM

Couple of problems in this suggested system:

1) We know from the historical experience of both faction play AND solaris that once make top tier players fight one another, they just stop playing the game. If they can't seal club, they stop playing that game mode.

2) As soon as you stack teams with only 95th percentile players (or any group of higher end players), their individual match scores will usually start dropping because it's not possible to farm damage and kills the way it is in games with a mixed grouping of skill levels. As noted above, the very best players will hit CT and head most of the time, perversely leading to lower match scores the more their skill levels rise. Does the system then bump them temporarily down to lower tiers again, creating a never ending cycle where almost no one is 95th percentile for very long?

#48 Eatit

    Member

  • PipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 286 posts

Posted 26 April 2019 - 11:08 AM

Wow 5th Fedcom Rat,

1) That's a pretty ugly thing to say about top tier players. I don't think it's true. In the Comp scene they have leagues with tiers. They don't pit the best against the worst. In FW that's a different story as there is no skill based match making. I think having a skill based match maker is what most players want including the top tier people. There may be some outliers but I don't think they're the norm. Also if they already left because they can't seal club then good we are better off without them.

2) I agree that the top players won't be able to pad stats by farming damage. They probably will have their skill/MS drop as a result. Players in lower tiers will still be able to do this. I guess the people saying Match Score needs to be looked at and values placed accordingly are right.

#49 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 26 April 2019 - 11:14 AM

View Post5th Fedcom Rat, on 26 April 2019 - 10:02 AM, said:

Couple of problems in this suggested system:

1) We know from the historical experience of both faction play AND solaris that once make top tier players fight one another, they just stop playing the game. If they can't seal club, they stop playing that game mode.

2) As soon as you stack teams with only 95th percentile players (or any group of higher end players), their individual match scores will usually start dropping because it's not possible to farm damage and kills the way it is in games with a mixed grouping of skill levels. As noted above, the very best players will hit CT and head most of the time, perversely leading to lower match scores the more their skill levels rise. Does the system then bump them temporarily down to lower tiers again, creating a never ending cycle where almost no one is 95th percentile for very long?


No, it'll create a system where people that take low value high match score actions to get higher match scores will be placed on teams where they are expected to perform the same as players that take high value low MS actions, and therefore the poor match making will continue since the MM continues to be unable to estimate a player's contribution to winning and losing.

Statistics is what I do for a living, I use a number of factors such as demographics, lab tests, vital signs etc to estimate whether one drug is better or another. Something like that needs to be done in MWO if you want the MM to be better (replacing what I listed with KMDD, solo kills, kill efficiency etc), but at this late stage in MWO's life span it's not worth the effort.

Edited by Nightbird, 26 April 2019 - 11:15 AM.


#50 PhoenixFire55

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 5,725 posts
  • LocationSt.Petersburg / Outreach

Posted 26 April 2019 - 11:35 AM

View Post5th Fedcom Rat, on 26 April 2019 - 10:02 AM, said:

1) We know from the historical experience of both faction play AND solaris that once make top tier players fight one another, they just stop playing the game. If they can't seal club, they stop playing that game mode.

And here I was wondering why you only play like 30 games per season lol

Edited by PhoenixFire55, 26 April 2019 - 11:36 AM.


#51 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 26 April 2019 - 12:06 PM

View PostNightbird, on 26 April 2019 - 11:14 AM, said:

No, it'll create a system where people that take low value high match score actions to get higher match scores will be placed on teams where they are expected to perform the same as players that take high value low MS actions, and therefore the poor match making will continue since the MM continues to be unable to estimate a player's contribution to winning and losing.

Statistics is what I do for a living, I use a number of factors such as demographics, lab tests, vital signs etc to estimate whether one drug is better or another. Something like that needs to be done in MWO if you want the MM to be better (replacing what I listed with KMDD, solo kills, kill efficiency etc), but at this late stage in MWO's life span it's not worth the effort.


Since you put it that way, the only thing that actually matters then in the long run is the ability to win (i.e. W/L ratio). As such, even a mascot that has a high W/L must be doing something right.



View PostEatit, on 26 April 2019 - 11:08 AM, said:

I guess the people saying Match Score needs to be looked at and values placed accordingly are right.


There is no need. See above.

Why complicate things?

Edited by Mystere, 26 April 2019 - 12:14 PM.


#52 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 26 April 2019 - 12:41 PM

View PostMystere, on 26 April 2019 - 12:06 PM, said:

Since you put it that way, the only thing that actually matters then in the long run is the ability to win (i.e. W/L ratio). As such, even a mascot that has a high W/L must be doing something right.


W/L is a factor, but there are other factors as well and the science is in how much to weigh each of them.

#53 Nesutizale

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Privateer
  • The Privateer
  • 3,239 posts

Posted 27 April 2019 - 12:23 AM

With just W/L a person who plays bad but gets lucky on team composition gets currently dragged along much more, rising in ranking, then he will loose as soon as he gets a bad team.
Result is and has been mention several times, that people end up in a Tier where they don't belong for their own good.

Just W/L is a good system for 1:1 games like Solaris 1:1 or Chess...thats why they use the ELO system for that.

#54 MrMadguy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,193 posts

Posted 27 April 2019 - 03:52 AM

I also have strong suspicion, that when matchmaking is bad, PGI also provide "bad" big open maps for voting in order to stretch match time,cuz otherwise it would be instant stomp.

Edited by MrMadguy, 27 April 2019 - 03:53 AM.


#55 Sjorpha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,472 posts
  • LocationSweden

Posted 27 April 2019 - 03:52 AM

View PostEatit, on 26 April 2019 - 08:43 AM, said:

I like your chart McGoat but the number of tier 1 players on there is pretty small. How will they get a decent wait time with such a low population?


The matchmaker doesn't necessarily have to separate different tiers into different matches. It only has to make sure there is close to equal total of skill on each team, so there should about as many players of each tier on each team.

I think with such low population, there is a strong argument for abandoning the idea of separation by tier altogether, allow whatever spread of skill there currently is in the queue, and just balance the total strength of the teams. It's not really a problem for tier 1 and tier 5 players playing together as long as the other team also have a similar spread.

#56 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 27 April 2019 - 04:13 AM

View PostSjorpha, on 27 April 2019 - 03:52 AM, said:


The matchmaker doesn't necessarily have to separate different tiers into different matches. It only has to make sure there is close to equal total of skill on each team, so there should about as many players of each tier on each team.


Yeah it's really not a hard concept to grasp as I see it.

Many seem to be unable to understand it however or simply don't understand how it works right now.

The entire point is exactly as I said a couple pages back. 190 Avg Match score and you make it into T1? That's the problem.

Edited by justcallme A S H, 27 April 2019 - 04:14 AM.


#57 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 27 April 2019 - 06:18 AM

View PostNesutizale, on 27 April 2019 - 12:23 AM, said:

With just W/L a person who plays bad but gets lucky on team composition gets currently dragged along much more, rising in ranking, then he will loose as soon as he gets a bad team.
Result is and has been mention several times, that people end up in a Tier where they don't belong for their own good.

Just W/L is a good system for 1:1 games like Solaris 1:1 or Chess...thats why they use the ELO system for that.


Over the long term? It's highly unlikely unless a person is really lucky. A very lucky person is an asset to the team. Posted Image


View PostNightbird, on 26 April 2019 - 12:41 PM, said:


W/L is a factor, but there are other factors as well and the science is in how much to weigh each of them.


You say "science", I say "dimensionality reduction". Posted Image

Edited by Mystere, 27 April 2019 - 06:19 AM.


#58 LowSubmarino

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • 1,091 posts

Posted 27 April 2019 - 06:32 AM

Just had a match on mining, we lost 0:12.

Complete masscare.

Standard approach center ramp to then stop, block each other, swarm to each side and autmatically split and then get completly annihilated - team.

Id put each and every single one of them in Tier 5.

#59 Anthrax87

    Member

  • Pip
  • The 1 Percent
  • The 1 Percent
  • 15 posts

Posted 27 April 2019 - 06:48 AM

I don't think any of this is the answer, cause a players match score on assualts would be very difference to that on lights.

It should do this.
1> Pair 24 players in a lobby
2> Determine their match score for the chassis type (ligh,med,heavy,assault) they're running in the game
3> Try to balance the teams by this TOTAL match score (not average, total, so if a team has what i like to call a Tier 0 player, they can be paired with ******** players)
4> Then try to balance the mechs by tonnage

EZ

#60 LowSubmarino

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • 1,091 posts

Posted 27 April 2019 - 07:14 AM

View PostAnthrax87, on 27 April 2019 - 06:48 AM, said:

I don't think any of this is the answer, cause a players match score on assualts would be very difference to that on lights.

It should do this.
1> Pair 24 players in a lobby
2> Determine their match score for the chassis type (ligh,med,heavy,assault) they're running in the game
3> Try to balance the teams by this TOTAL match score (not average, total, so if a team has what i like to call a Tier 0 player, they can be paired with ******** players)
4> Then try to balance the mechs by tonnage

EZ


Not a good idea.

No matter how good a player is, if you have completly inexperienced players that make basic mistakes (run into entire death ball of team red with like 4 heaveis and assaults) then the game is mostly lost right then and there. If youre down 3, 4 and then 5 mechs in seconds only a miracle comeback can save you and that doesnt happen that often. Maybe everyr 20th or 40th match.

A single very good player, even emp players or top team players, cant compensate that on any level.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users