Jump to content


The long and the short of skill-based matchmaking. why it's not that simple and yet better than ...

Matchmaker matchmaking

  • Please log in to reply
139 replies to this topic

CamperKaempfer #1 Posted 04 March 2018 - 06:59 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

FOR THOSE WHO COMPLAIN THAT THEIR TEAM IS ALWAYS THE WORSE ONE, ONE THING YOU NEED TO KNOW BEFORE YOU GO ON READING

 

while it's totally possible that a skill-based matchmaking can provide a better experience for both good and bad players, and the more so the further they stray from the average. it would NOT make you win more, unless you're a bad player yourself, and this is why:

 

excluding yourself, averagely, over a large enough number of games, your team is as good as any other team. sometimes it's stronger than the enemy, sometimes it's weaker. the only constant is you. if you're better than average so will be, on average, your team.

 

complaining about me and my terrible stats is illogical because i'll be your enemy more often than i'll be your ally, because the enemy team has one more place available (since i can't be you, your team only has 14 places to offer me, while the enemy's has 15)

 

now, let's get started

 

 

SKILL- BASED MATCHMAKING IS INCOMPATIBLE WITH WIN-RATIO AS A MEASUREMENT OF SKILL

 

skill-based matchmaking doesn't pair well with win ratio as a system of measuring skill. bad players such as myself would play against equally bad players and therefore would no longer end up in the bottom, while the better you become, the better the players you're pitted against will be and therefore you won't perform as well.

 

if you want a skill based matchmaking you need it to be based on something other than win-ratio. average experience, however it's calculated, can be a good candidate, but i want to talk about a way to account for victories and defeats instead. more on that later.

 

otherwise, no matter how good you are, your win ratio will always be about 50% and there is only one way around it, which won't take you very far anyway:

 

you could "cheat" the system by improving constantly (presumably after starting off poorly), since the matchmaker can only know about your performance so far.

 

the way this would work is that you can trick the matchmaker into thinking you're this good and deploy you against players that are also this good, while, instead, you'll be playing slightly better than that every time.

 

but, at some point you would still reach your actual skill level, and defeats will stack up just as fast as victories. for an infinite number of battles, the limit tends to a ratio of 1:1 and the victories you collected when the matchmaker thought you were worse than you actually are become less and less significant.

 

 

CUMULATIVE SCORE SYSTEM

 

a matchmaker that accounts for skill would have to jettison win ratio as a measurement of skill in favor of a cumulative "score", and this is exactly what games that employ as skill-based matchmaking, as well as many that don't, do

 

online chess games are a good example of both cumulative score and skill-based matchmaking. in those games your stats are just one number known as score, which is similar with world of tanks' ranked battles' chevron system, but with one key difference:

 

the higher the score of your opponent compared to yours, the more your score will increase in case of victory, and the less it will decrease in case of defeat.

 

similarly, the lower the score of your opponent is compared to yours, the fewer points you'll be adding to your score in case of victory, but the more you'll lose in case of defeat.

 

this way, unlike chevrons, your score won't rise indefinitely, because the many victories you'll achieve against the worse majority or players will be balanced by the fewer defeats you'll be inflicted by the minority of players that are only marginally better than you.

 

still, players with different levels of skill would possess different scores.

 

this score can have two purposes:

1) to replace or supplement win-ratios as player stats, allowing players to recover quickly from a long losing streak, such as that many players experience when start playing. this way won't be forever affected by the increasingly lighter but never vanishing toll of your noob days and your stats will be a better and constantly updated reflection of your skill level

2) to serve as an indicator (either hidden or not) to by used by a skill-based matchmaking that accounts for how often a player wins rather than how much experience they earn on average (whether that's preferable i'll leave to you to decide)

 

both usages are at odds with neither each other nor the display of any kind of stats that is currently being used

 

 

HOW IT WOULD WORK:

 

in online chess games you can usually select the maximum difference between your score and that of potential opponents. set it low and you'll only play against adversaries that are neither much better not much worse than yourself, and your win ratio (if it were recorded alongside your score) would be about 50%. set it infinite (effectively skill factor from the matchmaking and making it simply random) and it becomes like the chevron-neutral ranked battles in world of tanks, exactly as i described it before and with the same difference i mentioned before.

 

PS: any kind of skill-based matchmaking for ranked battles is still a terrible idea, because it would simply eliminate the influence of skill and everyone would be earning about the same number of chevrons, which would be 0 if the system is chevron neutral.

 


Edited by CamperKaempfer, 05 March 2018 - 12:12 AM.


jabster #2 Posted 04 March 2018 - 07:07 PM

    Field Marshal

  • Beta Tester
  • 12598 battles
  • 24,072
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostCamperKaempfer, on 04 March 2018 - 06:05 PM, said:

next time i write something important i must remember write the title in bold

 

Or even better use the pinned thread.

CamperKaempfer #3 Posted 04 March 2018 - 07:12 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

View Postjabster, on 04 March 2018 - 07:07 PM, said:

 

Or even better use the pinned thread.

 

thanks. done it

Long_Range_Sniper #4 Posted 04 March 2018 - 07:18 PM

    General

  • Player
  • 33446 battles
  • 9,158
  • [-MM] -MM
  • Member since:
    04-04-2011

Moved to pinned ;-)


Edited by Long_Range_Sniper, 04 March 2018 - 07:18 PM.


OreH75 #5 Posted 04 March 2018 - 07:20 PM

    Captain

  • Player
  • 48853 battles
  • 2,295
  • [OXIDE] OXIDE
  • Member since:
    05-29-2013
Now try to implement chess MM when you play 15 vs 15 randomly selected players and everyone plays one chess-piece on the same board independently of the others players on his team..good luck skill balancing that chess game.

evilchaosmonkey #6 Posted 04 March 2018 - 07:25 PM

    Lieutenant

  • Player
  • 16718 battles
  • 1,756
  • [EIGHT] EIGHT
  • Member since:
    05-04-2013

Spectacles, testicles, wallet and watch.

 

Dear lord no thanks.

 

Something easier, like I don't know simply balance the green team and red team first, that maybe a better place to start.



CamperKaempfer #7 Posted 04 March 2018 - 07:33 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

View PostOreH75, on 04 March 2018 - 07:20 PM, said:

Now try to implement chess MM when you play 15 vs 15 randomly selected players and everyone plays one chess-piece on the same board independently of the others players on his team..good luck skill balancing that chess game.

 

that chess game is a mess regardless of whether it's balanced. but if you want to balance it you jsut need to pick players with similar levels of expertise. i dont' understand what your point is

Edited by CamperKaempfer, 04 March 2018 - 08:03 PM.


OreH75 #8 Posted 04 March 2018 - 07:42 PM

    Captain

  • Player
  • 48853 battles
  • 2,295
  • [OXIDE] OXIDE
  • Member since:
    05-29-2013

View PostCamperKaempfer, on 04 March 2018 - 07:33 PM, said:

 

they wouldn't be randomly selected. they would all have similar score, and their score would rise or decrease more after a victory or a defeat based on the difference between the individual score and the average of the battle. i don't see why it can't be done. alternatively, there's experience as i mentioned

 

Join a clan and fight some advances against similar skilled clans and tell me how many of them end in 15-54321?

Shivva #9 Posted 04 March 2018 - 07:44 PM

    Lieutenant

  • Player
  • 29809 battles
  • 1,958
  • [J_A_G] J_A_G
  • Member since:
    05-20-2012

View PostCamperKaempfer, on 04 March 2018 - 05:59 PM, said:

 

 

a skill-based matchmaking can provide (in my opinion) a better experience for both good and bad players

 

Why should bad players get a better experience for being bad?..there would zero motivation for them to improve if both good and bad players both benefit. Let good players benefit and bad players suffer until they learn to improve.


Edited by Shivva, 04 March 2018 - 07:44 PM.


vasilinhorulezz #10 Posted 04 March 2018 - 07:45 PM

    Second Lieutenant

  • Player
  • 22813 battles
  • 1,109
  • Member since:
    09-26-2014
People really need to go back to school and read what "random" means.

m1x_angelico #11 Posted 04 March 2018 - 07:52 PM

    Warrant Officer

  • Player
  • 23153 battles
  • 878
  • [-VETO] -VETO
  • Member since:
    01-04-2015

Skill based MM can be made quite easily. Anyone saying it can't be done, either doesn't have sufficient mental capacity to comprehend it or has some sort of agenda.

 

Would it be "perfect" up to certain people meaning of perfection - no. Then again, perfection is just a subjective abstract concept.

Would it make the random games much better and enjoyable - yes.

Will WG introduce it -  most likely no. (for reasons of this conclusion, a quick search of the forum will yield numerous treads and posts about most likely reasons behind this).



_Anarchistic_ #12 Posted 04 March 2018 - 07:54 PM

    Second Lieutenant

  • Player
  • 41046 battles
  • 1,081
  • Member since:
    01-07-2015

makes no difference

 

playing ranked and its prtty much the same at every rank

 

biggest issue is not imbalanced teams but imbalanced maps

 



CamperKaempfer #13 Posted 04 March 2018 - 08:00 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

View PostShivva, on 04 March 2018 - 07:44 PM, said:

Why should bad players get a better experience for being bad?..there would zero motivation for them to improve if both good and bad players both benefit. Let good players benefit and bad players suffer until they learn to improve.

 

because all of the complains i noticed come from (allegedly) good players bitching over their terrible team. i was being condescending. plus being enjoyable is the whole point of a game. anyway i dont' really have a problem with the current matchmaking. i wan't pushing an agenda, i was just providing ideas and alalyses
 

View Postvasilinhorulezz, on 04 March 2018 - 07:45 PM, said:

People really need to go back to school and read what "random" means.

 

edited. now it's more clear
 

Shivva #14 Posted 04 March 2018 - 08:31 PM

    Lieutenant

  • Player
  • 29809 battles
  • 1,958
  • [J_A_G] J_A_G
  • Member since:
    05-20-2012

View PostCamperKaempfer, on 04 March 2018 - 07:00 PM, said:

 

because all of the complains i noticed come from (allegedly) good players bitching over their terrible team. i was being condescending. plus being enjoyable is the whole point of a game. anyway i dont' really have a problem with the current matchmaking. i wan't pushing an agenda, i was just providing ideas and alalyses

 

Nothing wrong with terrible teams, terrible enemy team increases chance of win for my team. Terrible friendly team means I 'should' be able to contribute more than I normally do....either way both situations can end in me having a fun game.

CamperKaempfer #15 Posted 04 March 2018 - 08:49 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

View PostShivva, on 04 March 2018 - 08:31 PM, said:

 

Nothing wrong with terrible teams, terrible enemy team increases chance of win for my team. Terrible friendly team means I 'should' be able to contribute more than I normally do....either way both situations can end in me having a fun game.

 

apparently not everyone thinks the same way as you and my post is directed mostly at them. i wanted to present possibilities for a sbmm to those who want one, rather than win over those who don't. as i said, i don't really have a problem with the current matchmaking. i just wanted to clear the waters
 

View PostOreH75, on 04 March 2018 - 07:42 PM, said:

 

Join a clan and fight some advances against similar skilled clans and tell me how many of them end in 15-54321?

i don't understand you point. some people want a sbmm and i provided some ideas on how it could work. and why's there a question mark? is that an order or a question?

View PostOreH75, on 04 March 2018 - 07:20 PM, said:

Now try to implement chess MM when you play 15 vs 15 randomly selected players and everyone plays one chess-piece on the same board independently of the others players on his team..good luck skill balancing that chess game.

that chess game is a mess regardless of whether it's balanced. but if you want to balance it you jsut need to pick players with similar levels of expertise. i dont' understand what your point is


 

Edited by Daxeno, 22 May 2018 - 05:12 PM.


Balc0ra #16 Posted 04 March 2018 - 09:02 PM

    Field Marshal

  • Player
  • 67361 battles
  • 17,081
  • [WALL] WALL
  • Member since:
    07-10-2012

The only thing you needed to do vs writing all of that. Is simply say how well it worked for that other game. They got it because they players wanted it really badly. And theirs worked a bit to well and killed off PVP more or less in a few weeks. Well it was one of a few handful of reasons. But that was the last straw for most. Those who where used to a high WR, suddenly got a 48 to 52% daily WR. Not 60%. So they left. As their WR still dropped even with skilled based MM. Imagine that. So if WG adds it. Just be sure it don't work to well.

 

Also it never removed lemming fails, camping fails or anything. It was all still there. As that's not related to low skill players only. Contrary to popular belief.



Archaean #17 Posted 04 March 2018 - 09:04 PM

    Second Lieutenant

  • Player
  • 15663 battles
  • 1,242
  • [S4LT] S4LT
  • Member since:
    04-25-2015

View PostCamperKaempfer, on 04 March 2018 - 06:59 PM, said:

 

don't complain about me and my terrible stats because i'll be your enemy more often than i'll be your ally, because the enemy team has one more place available (since i can't be you, your team only has 14 places to offer me, while the enemy's has 15)

 

massive threat me big scare



CamperKaempfer #18 Posted 04 March 2018 - 09:29 PM

    Staff Sergeant

  • Player
  • 23461 battles
  • 369
  • Member since:
    01-10-2017

View PostArchaean, on 04 March 2018 - 09:04 PM, said:

massive threat me big scare

 

you're so silly. that was the opposite of a threat. it means that other players being bad is good for you. try to understand before opening your mouth

xBiggiex #19 Posted 04 March 2018 - 09:35 PM

    Private

  • Player
  • 36552 battles
  • 31
  • [LCSGY] LCSGY
  • Member since:
    07-20-2013

LOOOL, skill is who can pay more this days. So in that line of idea MM has nothing to do with win/skill and what so ever. You pay you are good you dont you are there to feed those who pay.

 



m1x_angelico #20 Posted 04 March 2018 - 10:55 PM

    Warrant Officer

  • Player
  • 23153 battles
  • 878
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View PostxBiggiex, on 04 March 2018 - 09:35 PM, said:

LOOOL, skill is who can pay more this days. So in that line of idea MM has nothing to do with win/skill and what so ever. You pay you are good you dont you are there to feed those who pay.

 

 

So, if I understand correctly you consider yourself "food" for others? That's interesting...





Also tagged with Matchmaker, matchmaking

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users