Jump to content


Skill Based Matchmaking: Why Not?

Matchmaking

  • Please log in to reply
36 replies to this topic

Bavor #21 Posted Dec 09 2017 - 23:50

    Major

  • Players
  • 33935 battles
  • 3,170
  • [REL-A] REL-A
  • Member since:
    04-21-2013

In the past I've said we don't need skill based matchmaking that equals the skill between the teams exactly, but a matchmaker that limits the skill difference between teams.  When I looked at data from a few thousand replays, if the difference in average WN8 between the teams is larger than a certain percentage, the team with the higher average WN8 will win 90-95% of the time.  The higher the tier, the higher the win percentage is for the team with the higher WN8.  When you get into a difference of average WN8 being greater than 20% its a huge jump in win rate for the team with the higher WN8.

 

Overall, for the average player, the matchmaking equals out to average, but about 15% of the battles you will win no matter how bad you play and about 15% of the battles you will lose no matter how good you play due to skill imbalance between teams.  The data I've seen from players who are in the top 0.01% of players(by WN8) seems to support this.  A platoon of 3000+ WN8 players seems to max out at about 85% win rate when they play together over 100+ battles.  Getting rid of those 30% of battles where the results are almost predetermined would be an improvement for most players. 

 

The only objection I've seen to limiting the skill difference between teams is among the top 0.1% of players who said it may lower their recent win rates 1-3% because they would get less battles where there is a huge skill advantage in their favor due to the randomness of matchmaker.  They would get less battles with a large number of sub 450 WN8, sub 46% win rate players on the enemy team.

 

With the current 3/5/7 matchmaking, the team with the better top tier players also is much more likely to win, especially in tiers 7 through 10.  It is frustrating being bottom tier and doing more damage than all of your team's top tier tanks and losing because the players in top tier tanks don't have the skill to help their team win.

 

Wargaming could limit the skill difference between teams based on WoT Personal Rating and eliminate the 30% of battles that are almost automatic wins or automatic losses.



Bavor #22 Posted Dec 09 2017 - 23:54

    Major

  • Players
  • 33935 battles
  • 3,170
  • [REL-A] REL-A
  • Member since:
    04-21-2013

View PostHorribad_At_Tanks, on Dec 09 2017 - 16:41, said:

 

I always focus the guds when playing arty because they have the most influence on the match outcome. Why waste shots on the tomato in the open when I can wait for the gud to poke out and smack him back making him retreat and take him out of the fight? 

 

The problem with that method is the best way to defeat the top players on the enemy team is to take out all of their support.  Remove the lower skill players who spot for them take shots for them, provide cover for them.  Before I started playing arty for personal missions, I had a 58% win rate solo pub in tier 8 arty because I didn't focus the top players.  Instead I did as much damage as possible where my team needed the support instead of focusing the top 3 players by WN8 or win rate.

 

The problem is 95% of pub battle arty players are basically useless because they don't think about where or who they are shooting and they rarely or never relocate for better shots.



SupaD63 #23 Posted Dec 10 2017 - 00:05

    Corporal

  • -Players-
  • 33159 battles
  • 78
  • [TEX_2] TEX_2
  • Member since:
    02-08-2015

View PostPipinghot, on Dec 09 2017 - 16:54, said:

You are a cancer and I hope WG bans you.

 

This

shiil #24 Posted Dec 10 2017 - 00:12

    Staff sergeant

  • Players
  • 50587 battles
  • 449
  • [DV] DV
  • Member since:
    06-26-2011

View PostPipinghot, on Dec 09 2017 - 21:54, said:

You already know the answer, now you need to use your brains and think about why that would be a problem. If you want to play a game where you don't have to work to win, then go play a single-player game. This is a competition game.

You are a cancer and I hope WG bans you.

"This is a competition game"  where is the competition when most battles end in 4 minutes from the start,  15-3, 15-4 or worse 15 -1  results are not competition in my opinion. In one game you are chasing to make some damage while your team is wiping out the enemy in 3 to 4 minutes, then the next game   your team collapses in 3 minutes and you barely fire 3-4 shots ( unless you are in those fast meta medium tanks) before you get surrounded and get smoked. 

 



moon111 #25 Posted Dec 10 2017 - 00:13

    Major

  • Players
  • 32346 battles
  • 2,528
  • Member since:
    06-29-2013

Match-making and RNG are part of the equation used to balance out your performance average.  I've never witnessed such bad match-making as when the enemy team has a really bad player in a KV-1.

The game will practically rig the game so that player can get a 48% win-rate.  Have an incredible night with a 80% win-rate... it isn't just the law of averages that has my XVM showing a ton of red players

on my team and blue and purple's on the enemy team the next night.  There's skilled based MM, it's just designed to squeeze you to the middle of the pack.



Silky_Slim #26 Posted Dec 10 2017 - 00:31

    Staff sergeant

  • -Players-
  • 14296 battles
  • 260
  • [AVNGE] AVNGE
  • Member since:
    12-05-2015

This may sound a bit conspiratorial so put on your tin foil hats, but you have to really suck to have less than a 45% WR and really good to have 55-60+% WR. As it has been stated by many, in this post and others, MM is balanced to give and take in the favorable and unfavorable matches an individual see.  Here comes the conspiracy. So if matchmaking is controlled in that way, a side affect or expected outcome, would be more matches which increases gold and credit expenditure by player base.



GeorgePreddy #27 Posted Dec 10 2017 - 01:15

    Major

  • Players
  • 14680 battles
  • 13,065
  • Member since:
    04-11-2013

1. It doesn't matter whether you understand why SBMM is a terrible thing for Random Battles or not, because... WG knows and that's all that matters really.

 

2. Doesn't deserve an answer and you know it.



GeorgePreddy #28 Posted Dec 10 2017 - 01:16

    Major

  • Players
  • 14680 battles
  • 13,065
  • Member since:
    04-11-2013

View Postshiil, on Dec 09 2017 - 20:12, said:

 most battles end in 4 minutes from the start

 

 

No, they don't.

 

According to vBAddict.net, the average battle length for the tank that has the SHORTEST battles, the Pz. S35 739 (f), is 5.37 minutes.

 

The average battle length for the tank that has the LONGEST battles, the T110 E3, is 7.29 minutes (almost double your figure).

 

(Don't look now but your hyperbole is showing)



strenfoo #29 Posted Dec 10 2017 - 02:55

    Captain

  • -Players-
  • 12425 battles
  • 1,256
  • Member since:
    08-15-2015

View PostWinningWithWarchild, on Dec 09 2017 - 14:41, said:

Remind me again why it would be a bad idea to try and have even teams so there's an equal chance of winning every game?  I honestly cannot remember the logic behind the current system.  Something about we'd all have 50% winrates or something like that.  I really need a refresher on the reasoning and why so many people are in support of it.

 

Why don't you just read one of the thousands of other threads created over the years that discuss that exact topic in detail?

Horribad_At_Tanks #30 Posted Dec 10 2017 - 09:27

    Major

  • Players
  • 2268 battles
  • 6,263
  • Member since:
    11-07-2012

View PostBavor, on Dec 09 2017 - 17:54, said:

 

The problem with that method is the best way to defeat the top players on the enemy team is to take out all of their support.  Remove the lower skill players who spot for them take shots for them, provide cover for them.  Before I started playing arty for personal missions, I had a 58% win rate solo pub in tier 8 arty because I didn't focus the top players.  Instead I did as much damage as possible where my team needed the support instead of focusing the top 3 players by WN8 or win rate.

 

The problem is 95% of pub battle arty players are basically useless because they don't think about where or who they are shooting and they rarely or never relocate for better shots.

 

Well yes of course. Was simply saying that I ignore the tomato in the open to focus the gud player whenever that gud player starts to engage even though the tomato is easy damage. Otherwise I'm looking for low health tanks to take out of the fight or some other tank that needs some arty lovin like a superheavy that starts to push and needs some tracks taken out. I don't ignore the rest of the enemy just to single out the gud player otherwise I wouldn't get any large amount of damage or kills.



Pipinghot #31 Posted Dec 10 2017 - 17:15

    Major

  • Players
  • 25960 battles
  • 10,169
  • [IOC] IOC
  • Member since:
    11-20-2011

View Postshiil, on Dec 09 2017 - 18:12, said:

View PostPipinghot, on Dec 09 2017 - 21:54, said:

You already know the answer, now you need to use your brains and think about why that would be a problem. If you want to play a game where you don't have to work to win, then go play a single-player game. This is a competition game.

You are a cancer and I hope WG bans you.

"This is a competition game"  where is the competition when most battles end in 4 minutes from the start,  15-3, 15-4 or worse 15 -1  results are not competition in my opinion. In one game you are chasing to make some damage while your team is wiping out the enemy in 3 to 4 minutes, then the next game   your team collapses in 3 minutes and you barely fire 3-4 shots ( unless you are in those fast meta medium tanks) before you get surrounded and get smoked.

When two teams of professional athletes compete and one team wins 30-0, that is still a competition, it's just that one team severely outperformed the other.

When two professional boxers meet and there is a knockout in the first round, that is still a competition, it's just that one boxer severely outperformed the other.

 

The duration of a battle means nothing in terms of it being a competition.

 

Furthermore, a 50-50 battle can end in 4 minutes, and can end |15-3, 15-4 or worse". You are confusing the difference between "a competition" and "a competition in which you, the player, are guaranteed that the battle will play out exactly the way you want it to."

 

If you don't like getting tackled, don't play football.

If you don't like getting punched, don't become a boxer.

If you don't understand why some battles end quickly or in a large margin of victory, then don't play WoT (or even better, learn the game better so you understand why those things happen.) No one says you're forced to like WoT, if you like other games better that's your right, but you're failing to understand how and why this game works the way it does.

 

You may not like how WoT battles play out, but it's still a competition game, and that means players beating other players as quickly and efficiently as they can. If you don't like getting left behind during victories, and if you don't like your team getting wiped out before you can be useful in the battle, then learn what you need to do so that you can be more effective and help your teams more.


Edited by Pipinghot, Dec 10 2017 - 17:15.


__WarChild__ #32 Posted Dec 10 2017 - 19:44

    Major

  • -Players-
  • 30789 battles
  • 5,351
  • [OPIC] OPIC
  • Member since:
    06-03-2017

View PostPipinghot, on Dec 09 2017 - 15:54, said:

You are a cancer and I hope WG bans you.

I was joking.

View PostSupaD63, on Dec 09 2017 - 17:05, said:

 

This

You're an idiot.

View PostGeorgePreddy, on Dec 09 2017 - 18:15, said:

2. Doesn't deserve an answer and you know it.

He knows I was joking.

 

:trollface:



LpBronco #33 Posted Dec 11 2017 - 01:51

    Major

  • Beta Testers
  • 43817 battles
  • 2,774
  • [TG] TG
  • Member since:
    11-19-2010

View PostGurdy2383, on Dec 09 2017 - 17:05, said:

 

My previously posted thoughts on SBMM (Skill Based Match Making): In this I'm referring to a system where the MM would pick 30 players first, and then balance the teams equally from there.

 

1.) First of all, I would argue that a Skill Based MM system would be largely ineffective to solve the so-called problems that people want solved in the first place. This

     system would rarely result in balanced teams. Things that would throw it off are:

 

*  3 man unicum platoon on one side, vs. a 3 man tomato platoon on the other team.

*  The unicums on one team are high tier, while the unicums on the other team are bottom tier.

*  If based on lifetime stats: One team gets a bunch of yellows that have recent blue stats, while the other team gets a bunch of yellows with recent yellow stats.

*  If based on recent stats: One team gets a a bunch of blues who have green recent stats because they've been grinding things that bring their stats down, while the other side
   gets a bunch of true greens.

 

Re-rolls vs. true lifetime stats, good players in Chi-ri's vs. good players in T-54's, players that have good stats but are just starting to play higher tiers, stock tanks, etc. would also add to the unlikelihood that this system would result in balanced teams.
 
2.) Putting any type of skill based MM into a game that has tech tree progression doesn't jive. That's even if it's a simple extra sorting of teams once the players have been
     chosen. The status quo is purely random therefore fair. Any messing with the randomness will inherently lead to some unfair distribution of credits and experience in many
     situations.

 

Example: The MM pulls 30 players 3 of which are purple, 4 green, 5 yellow, and 18 reds. Whatever metric the system uses to sort the players, it would always have to value purples way higher than even greens, because the gap between a purple and a green is much greater than the gap between a green and a yellow, etc. So:

Team 1 could expect a team that consists of 2 purple, 0 green, 2 yellow, & 11 reds.

Team 2 would then be 1 purple, 4 green, 3 yellow, & 7 reds.

 

Maybe it would be a little different, but the theme would always be the same: the team with the most unicums will have the most tomatoes as well. These unicums will be asked to carry their teams to victory more often than other players will. They will therefore earn less credits and experience if they don't carry. The MM will consistently place tankers based on their stats into the same situations which either benefit their progression through the tech trees, or hurt it. This example doesn't even start to talk about how unfair a unicum placed on a team of tomatoes in a low tier vehicle will be. Remember that progression through the tech trees is the part of this game that is monetized. SBMM creates a situation where the monetization of the game is uneven for all players.

 

3.) This system will also encourage players to not improve. Whenever a player, for example, turns from green to blue they are instantly transformed into a player that has

      worse teammates with SBMM. There's no way around this. Improvement means that you will be responsible to carry your team more, and if you don't you won't make
      as many credits and experience as you used to. This is not a good way to encourage better play, and in fact is the opposite. This also ties into the fact that the fairness of
      the distribution of experience and credits is compromised in a in a tech tree progression style game when you add any skill based mm. There would
      undoubtedly be sandbaggers with SBMM that would be a big negative to gameplay.

 

4.) This system would also reduce the effect of one's own skill level on influencing their win rate. The status quo is purely random. This means that good players win more, while
      poor players win less. Average players will fall somewhere near 50% win rate. With SBMM the mm would always attempt to make the teams even. This way good
      players win about 50% of the time, poor players win about 50%, and average players will win 50% of the time. This is another side effect of your system that reduces the
      players' drive to improve their play. If they will win half of the time regardless of how good they are why care to improve? With your system if somebody starts winning more
      often than 50% of the time the mm will make sure to balance out that person's matches, and get them down to 50%. You're going from a random system that sometimes
      screws players out of winning to a structured system that is actually designed to constantly screw players over that start winning too much, and is designed to assist those
      who are winning too little.

 

SBMM wouldn't really result in much greater balance in matches. Steamrolls would still exist with nearly the same frequency as before. The MM would be powerless to sort through all of the variables that make up this game and it's player base. There is no good way to compare players apples-to-apples since this is a tech-tree centered game where everyone has different vehicles, crews, gold usage, tiers played, etc, etc, etc. SBMM would, though, create some new problems that are pretty unfair, and don't really make sense with the core premise of the game. This is not a good trade-off, and SBMM will do more harm than good the way I see it.

 

It's not that WG isn't listening, don't care, are stubborn, or think it would be too hard to implement. It's just that they've thought about it a lot more than you and I have, and have decided that it's not a good idea.

 

Here's the main reason why Skill Based MM will never happen, though:

 

WG doesn't want to go to the "Pay more money to make the game harder for yourself" model? Think about it:

 

Take, for example, Player A and Player B. Lets assume that they both are the exact same person with the same skill, learning curve, etc. They are exactly the same, except Player A decides on day one to buy $50 worth of gold. Now player A is running in matches with a 100% trained crew while Player B is running with sub-100% trained crews. In a short time, Player A's stats start to show that he is more effective than Player B. He misses less shots, he can afford full equipment for all of his tanks, he transfers crews from one vehicle to the next fully trained so he's starting to get some perks researched, he free xp's his way past stock modules, etc. Player A is now winning more than Player B. Then somebody tells Player A about the "skill based MM" in WoT. He sees that Player B is getting much easier matches than he is. Player A feels upset that he spent money to make the game harder for himself. Player A vows to never spend money on this game again.

 

Yeah, I can't see WG getting excited about that idea.

 

Thank you for taking the time to lay out the REASONS SBMM is a non-starter. The disparity in vehicles makes vehicle balancing the only reasonable way to achieve some balance and even then trying to balance by role is also too limiting and repetitious. The matchmaking before the templated change would have worked fine if they had included the special cases they now allow now.

xKGB_CIAx #34 Posted Jan 16 2019 - 20:38

    Corporal

  • -Players-
  • 44170 battles
  • 42
  • [CAMLT] CAMLT
  • Member since:
    06-22-2016

View PostBavor, on Dec 09 2017 - 23:50, said:

In the past I've said we don't need skill based matchmaking that equals the skill between the teams exactly, but a matchmaker that limits the skill difference between teams.  When I looked at data from a few thousand replays, if the difference in average WN8 between the teams is larger than a certain percentage, the team with the higher average WN8 will win 90-95% of the time.  The higher the tier, the higher the win percentage is for the team with the higher WN8.  When you get into a difference of average WN8 being greater than 20% its a huge jump in win rate for the team with the higher WN8.

 

Overall, for the average player, the matchmaking equals out to average, but about 15% of the battles you will win no matter how bad you play and about 15% of the battles you will lose no matter how good you play due to skill imbalance between teams.  The data I've seen from players who are in the top 0.01% of players(by WN8) seems to support this.  A platoon of 3000+ WN8 players seems to max out at about 85% win rate when they play together over 100+ battles.  Getting rid of those 30% of battles where the results are almost predetermined would be an improvement for most players. 

 

The only objection I've seen to limiting the skill difference between teams is among the top 0.1% of players who said it may lower their recent win rates 1-3% because they would get less battles where there is a huge skill advantage in their favor due to the randomness of matchmaker.  They would get less battles with a large number of sub 450 WN8, sub 46% win rate players on the enemy team.

 

With the current 3/5/7 matchmaking, the team with the better top tier players also is much more likely to win, especially in tiers 7 through 10.  It is frustrating being bottom tier and doing more damage than all of your team's top tier tanks and losing because the players in top tier tanks don't have the skill to help their team win.

 

Wargaming could limit the skill difference between teams based on WoT Personal Rating and eliminate the 30% of battles that are almost automatic wins or automatic losses.

 

ME I  CANT   SEE HOW ITS RANDOM AS THERES NO  WAY  TO  FLIP  A COIN  100  TIMES   AND NOT GET   45%  HEADS  JUST TODAY  40  GAME   ALL  VERY BAD TEAMS AND VERY  GOOD  RED TEAMS ARTYS THAT  SNAP  SHOT  YOU  AT FIRST SPOT  MY  ARTY  CANT HIT THE WALL  IN FRONT OF THEM GAME AFTER GAME  HEAVYS  THAT  CAMP  BASE TILL ITS WAY WAY TOO  LATE TO  ANY  THING   SCOUTS THAT    MOVE UP  NEXT TO ME   HEAVYS THAT HIDE BEHIND ME AND SHOOT GET   ME SHOT UNSPOTED THIS  IS OK  5 OUT OF 10  GAMES BUT  NOT 8.5 OUT OF 10  IM TRACKING TEAM PLAY TODAY    I  CALL bs rng   AND BS MM   FOR ME TODAY  EVERYTIME I  GET  MY  WINS UP  TO  ABOUT  59%  ITS LOSS WITH  NO  PINS NO SPOTING 5 OR 10  GAMES IN ROW 

DOMlNO #35 Posted Jan 16 2019 - 21:12

    Staff sergeant

  • Players
  • 17631 battles
  • 393
  • Member since:
    05-28-2012

View PostPipinghot, on Dec 10 2017 - 11:15, said:

You may not like how WoT battles play out, but it's still a competition game, and that means players beating other players as quickly and efficiently as they can.

 

This is EXACTLY why teams should be balanced, to make it COMPETITIVE!

 

Have you ever played a pick-up basketball game? You don't make teams by randomly choosing players, you pick teams to be as close as possible to make it a competitive game. Otherwise it wouldn't be any fun.



Silky_Slim #36 Posted Jan 16 2019 - 21:19

    Staff sergeant

  • -Players-
  • 14296 battles
  • 260
  • [AVNGE] AVNGE
  • Member since:
    12-05-2015

View PostxKGB_CIAx, on Jan 16 2019 - 20:38, said:

 

ME I  CANT   SEE HOW ITS RANDOM AS THERES NO  WAY  TO  FLIP  A COIN  100  TIMES   AND NOT GET   45%  HEADS  JUST TODAY  40  GAME   ALL  VERY BAD TEAMS AND VERY  GOOD  RED TEAMS ARTYS THAT  SNAP  SHOT  YOU  AT FIRST SPOT  MY  ARTY  CANT HIT THE WALL  IN FRONT OF THEM GAME AFTER GAME  HEAVYS  THAT  CAMP  BASE TILL ITS WAY WAY TOO  LATE TO  ANY  THING   SCOUTS THAT    MOVE UP  NEXT TO ME   HEAVYS THAT HIDE BEHIND ME AND SHOOT GET   ME SHOT UNSPOTED THIS  IS OK  5 OUT OF 10  GAMES BUT  NOT 8.5 OUT OF 10  IM TRACKING TEAM PLAY TODAY    I  CALL bs rng   AND BS MM   FOR ME TODAY  EVERYTIME I  GET  MY  WINS UP  TO  ABOUT  59%  ITS LOSS WITH  NO  PINS NO SPOTING 5 OR 10  GAMES IN ROW 

 

What? I'm picturing this in cut up magazine text. Ease off the caps and add in some punctuation. Now try to clearly state your point.

Altwar #37 Posted Jan 16 2019 - 21:39

    Major

  • Players
  • 58034 battles
  • 5,165
  • [-GNR-] -GNR-
  • Member since:
    04-24-2011

View PostxKGB_CIAx, on Jan 16 2019 - 11:38, said:

 

ME I  CANT   SEE HOW ITS RANDOM AS THERES NO  WAY  TO  FLIP  A COIN  100  TIMES   AND NOT GET   45%  HEADS  JUST TODAY  40  GAME   ALL  VERY BAD TEAMS AND VERY  GOOD  RED TEAMS ARTYS THAT  SNAP  SHOT  YOU  AT FIRST SPOT  MY  ARTY  CANT HIT THE WALL  IN FRONT OF THEM GAME AFTER GAME  HEAVYS  THAT  CAMP  BASE TILL ITS WAY WAY TOO  LATE TO  ANY  THING   SCOUTS THAT    MOVE UP  NEXT TO ME   HEAVYS THAT HIDE BEHIND ME AND SHOOT GET   ME SHOT UNSPOTED THIS  IS OK  5 OUT OF 10  GAMES BUT  NOT 8.5 OUT OF 10  IM TRACKING TEAM PLAY TODAY    I  CALL bs rng   AND BS MM   FOR ME TODAY  EVERYTIME I  GET  MY  WINS UP  TO  ABOUT  59%  ITS LOSS WITH  NO  PINS NO SPOTING 5 OR 10  GAMES IN ROW 

 

Uh oh.  I knew nothing good would come of the Chinese probe landing on the dark side of the moon.  They woke something up, and it is trying to communicate!







Also tagged with Matchmaking

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users