Jump to content


Proving SBMM Yields a 50% Win Rate


  • Please log in to reply
172 replies to this topic

DeviouslyCursed #1 Posted May 19 2019 - 17:59

    Staff sergeant

  • -Players-
  • 3220 battles
  • 381
  • Member since:
    12-13-2018

I already know my SBMM method would yield an eventual 50% win rate (using wins and losses to increase or decrease a rating that is used to place that person in a match). Some people have suggested other methods they think will lead to fair matches but not a 50% win rate. Personally, that doesn't make any sense, as keeping people who win more from continuing to win more is precisely the point of SBMM, but whatever. 

 

So, lets try a different approach. We'll use the a method of SBMM described in a previous post: Placing players of similar skill on opposite teams, balancing the teams. For instance, if 4 green players in a match, 2 green per side: 10 yellow, then 5 yellow per side, etc.

 

Now, some people claim that this won't screw over better players, because better players win more close games and that's how they have their high win rate. What is actually true is once a player has their high rating, they will win half of their 50% games, but they still win more games because due to random MM, they get more 51%+ chance to win games (it's because their high rating is being calculated into the chance to win). For some reason, it is difficult to convince people of this, and there is this continuous proclamation of: But good people win more battles when evenly matched, so therefore all the good players will still somehow keep winning more than 50%.

 

It's the equivalent of saying "Yeah, it's balanced, but then magic happens, and good people still win more, see?" Unfortunately, this can't happen. You see, if you have an equivalent player on the other side, then for each fight, there are two "players who are good and should win more" and one loses, the other wins (net of 50%). No matter how many battles you run, no matter how many are on each side, as long as it is balanced (same number on each side) some number of good players win, and an equal number of good players lose.

 

Match 1

Team1 vs Team 2

Team1: B G G G Y Y Y Y O O O R R R R

Team2: B G G G Y Y Y Y O O O R R R R

 

Team 1 wins

The result for wins/losses by skill gradient:

W/L

0/0

1/1

3/3

4/4

3/3

4/4

 

You can repeat that match as many times as you want. It won't matter. Regardless of who wins, the end result is the same. 50% win rate between the players of equal skill. This is true for 1 battle. It is true for 1k battles. What's worse, if someone from one of the skill brackets does manage to win more, they have pushed another player from that bracket BELOW 50%, which is completely insane if you think that is fair.

 

And BTW, it only gets worse when you consider that Draws are losses for both sides...

 

So, after 10 battles:

W/L

0/0

10/10

30/30

40/40

30/30

40/40

 

And as I already said, if they don't split those matches 50%, then 1 blue player is below 50%, 3 green players are below 50%, 4 red players are above 50%, etc.

 

And holy hell, it's doesn't matter if this is just "speculation." It is the application of logic, reason, and simple math.

 

I challenge anyone who thinks SBMM won't drive people to 50%, and/or screw over good players unjustly, to show teams and battle results that somehow show this! You have free reign to make team composition however you like and what ever team you want to win, as long as the teams are balanced. You can make different teams each match (still has to be balanced). Go ahead. Show me how this magic is supposed to work. Show me how you can place greens against greens, blues against blues, and somehow not end up with that category at 50% win rate (each player will either be 50% even, or some of those "good players who always manage to win more than they lose" will actual lose more than 50% so that another person in that category can win more than 50%).

 

Spoiler

 


Edited by DeviouslyCursed, May 19 2019 - 18:07.


SporkBoy #2 Posted May 19 2019 - 18:17

    Major

  • Players
  • 44684 battles
  • 2,259
  • [PZB] PZB
  • Member since:
    02-06-2014
Balance is a myth. Fair fights are a myth.

"Simple math" is at best a myth but in actuality code for "too stupid to understand" but really means you can't explain your method because it's fantasy.

There is plenty of data to work with  so where are all your scenario simulations proving your assertions (burden of proof is on you)?

Your furious hand waving and bold assertions do not constitute proof.

vonkevin #3 Posted May 19 2019 - 18:19

    Captain

  • Players
  • 30650 battles
  • 1,141
  • Member since:
    08-08-2011
And who is going to play a game where no-one is allowed to be good, and everyone is forced into being average?  Mediocrity is not the answer.

TankFullOfBourbon #4 Posted May 19 2019 - 18:23

    Major

  • Players
  • 29385 battles
  • 6,507
  • [DHO6] DHO6
  • Member since:
    08-10-2013
:popcorn:

DeviouslyCursed #5 Posted May 19 2019 - 18:24

    Staff sergeant

  • -Players-
  • 3220 battles
  • 381
  • Member since:
    12-13-2018

View PostSporkBoy, on May 19 2019 - 18:17, said:

Balance is a myth. Fair fights are a myth.

"Simple math" is at best a myth but in actuality code for "too stupid to understand" but really means you can't explain your method because it's fantasy.

There is plenty of data to work with so where are all your scenario simulations proving your assertions (burden of proof is on you)?

Your furious hand waving and bold assertions do not constitute proof.

 

Well, you're right, math that is simple for you may be a myth...



Rigour67 #6 Posted May 19 2019 - 18:27

    Staff sergeant

  • Players
  • 27009 battles
  • 327
  • [RIB] RIB
  • Member since:
    08-30-2013

Imagine playing a game in which, the better you get at the game, the worse your teammates become as some SBMM tries to "balance" the team with you on it with the team without you on it.

Conversely, the worse you play, the better your teammates become.

Now imagine what effect this has on toxicity in the player base.  Imagine a situation in which elite players see their teams full of tomatoes match after match, on purpose, as a direct result of the SBMM at work.

Exactly who would keep playing under those conditions of contest?

 


Edited by Rigour67, May 19 2019 - 18:28.


Treeburst #7 Posted May 19 2019 - 18:53

    Captain

  • Players
  • 31367 battles
  • 1,874
  • [HACKD] HACKD
  • Member since:
    04-20-2011
I don’t think you have a grasp on what the people that are proponents of this idea mean, or you do and are intentionally trying to obfuscate their argument.

Unless I am mistaking the argument entirely what the “new” proposal is not to form teams that are mirrored, but to shift players around that have already been “drafted” so to speak.

You will never get a truly mirrored match with the population NA has, blues and purples are too rare outside of tier 10. He’ll, for that matter greens might be too rare too.

As I said in an earlier post on this, people like myself would actually be helped by this change, since odds are I’m going to be the only good player on either team anyway (which can’t really be balanced out if you only have the 30 pre-selected players as a pool to pick from), and on the rare occasion there is a handful of good players they’ll be divided up equally. Platoons of good players will likely find a few games slightly harder, since having three good players on the team will pretty much ensure every other good player in the match will be on the opposing team.

owlgator #8 Posted May 19 2019 - 18:59

    Captain

  • Players
  • 15371 battles
  • 1,634
  • [W--W] W--W
  • Member since:
    05-27-2011

I'm confused by this topic.  Are there players who don't believe we'd see everyone at 50% over a large number of battles?

 

I don't think that's the hesitation (to put it mildly) for SBMM.  Instead of improving, do you really think below-average players should just get more favorable conditions in a game?  You're enabling their poor play, as they will just get rewarded for being below average.



Hellsfog #9 Posted May 19 2019 - 19:04

    Major

  • Players
  • 37626 battles
  • 6,122
  • [VILIN] VILIN
  • Member since:
    06-22-2011

View PostTreeburst, on May 19 2019 - 12:53, said:

I don’t think you have a grasp on what the people that are proponents of this idea mean, or you do and are intentionally trying to obfuscate their argument.

Unless I am mistaking the argument entirely what the “new” proposal is not to form teams that are mirrored, but to shift players around that have already been “drafted” so to speak.

You will never get a truly mirrored match with the population NA has, blues and purples are too rare outside of tier 10. He’ll, for that matter greens might be too rare too.

As I said in an earlier post on this, people like myself would actually be helped by this change, since odds are I’m going to be the only good player on either team anyway (which can’t really be balanced out if you only have the 30 pre-selected players as a pool to pick from), and on the rare occasion there is a handful of good players they’ll be divided up equally. Platoons of good players will likely find a few games slightly harder, since having three good players on the team will pretty much ensure every other good player in the match will be on the opposing team.

 

Treeburst, there are numerous threads on the forum explain why this new system won't work and isn't as simple as the proponents claim. For example, http://forum.worldoftanks.com/index.php?/topic/603970-skill-based-match-maker-versus-skill-balanced-match-maker/#topmost.  The fact is the OP is right. You and I won't benefit from it since we will be consistently given worse teammates to balance the teams. Yes, there is the rare occasion where it may help but on the majority of games, balance will require that we have the worse team.  

 

View Postowlgator, on May 19 2019 - 12:59, said:

I'm confused by this topic.  Are their players who don't believe we'd see everyone at 50% over a large number of battles?

 

I don't think that's the hesitation (to put it mildly) for SBMM.  Instead of improving, do you really think below-average players should just get more favorable conditions in a game?  You're enabling their poor play, as they will just get rewarded for being below average.

 

 

Yes, there are players who believe it. Yes, what they are hoping for is a mechanism that gives them more wins. 


Edited by Hellsfog, May 19 2019 - 19:08.


DeviouslyCursed #10 Posted May 19 2019 - 19:06

    Staff sergeant

  • -Players-
  • 3220 battles
  • 381
  • Member since:
    12-13-2018
 

View PostTreeburst, on May 19 2019 - 18:53, said:

I don’t think you have a grasp on what the people that are proponents of this idea mean, or you do and are intentionally trying to obfuscate their argument.

Unless I am mistaking the argument entirely what the “new” proposal is not to form teams that are mirrored, but to shift players around that have already been “drafted” so to speak.

You will never get a truly mirrored match with the population NA has, blues and purples are too rare outside of tier 10. He’ll, for that matter greens might be too rare too.

As I said in an earlier post on this, people like myself would actually be helped by this change, since odds are I’m going to be the only good player on either team anyway (which can’t really be balanced out if you only have the 30 pre-selected players as a pool to pick from), and on the rare occasion there is a handful of good players they’ll be divided up equally. Platoons of good players will likely find a few games slightly harder, since having three good players on the team will pretty much ensure every other good player in the match will be on the opposing team.

 

Actually, I think part of the problem is that everyone seems to have their own idea of what SBMM is. Then people discuss (and argue), while each person is talking about a different thing without realizing it.

 

This is why I defined it here, so at least we can talk about the same thing for a little bit.

 

What you just described would make really good people have all the worst people in a match on their team, every game. This is kind of what Rigour67 described above.


Edited by DeviouslyCursed, May 19 2019 - 19:06.


FastForward7 #11 Posted May 19 2019 - 19:10

    Staff sergeant

  • Players
  • 11090 battles
  • 498
  • Member since:
    03-31-2011

View PostSporkBoy, on May 19 2019 - 18:17, said:

Balance is a myth. Fair fights are a myth.
 

 

Yet WG's "balances" games for tank tier and type. But just for funsies, right? If you started seeing games with 6 arty, and 9 TD's (T7-T9) against 7 heavies and 8 meds (T8-T10) you'd soil your thong in your outrage. Go figure. 

FastForward7 #12 Posted May 19 2019 - 19:20

    Staff sergeant

  • Players
  • 11090 battles
  • 498
  • Member since:
    03-31-2011

View Postowlgator, on May 19 2019 - 18:59, said:

I'm confused by this topic.  Are their players who don't believe we'd see everyone at 50% over a large number of battles?

 

I don't think that's the hesitation (to put it mildly) for SBMM.  Instead of improving, do you really think below-average players should just get more favorable conditions in a game?  You're enabling their poor play, as they will just get rewarded for being below average.

 

Your post assumes that winning is all that matters. I mean, it is, as the game stands right now, but that's because that's how WG has set things up. If the game rewarded performance metrics instead of WR, things would be different. 

 

I'm sure some idiot will chime in any minute about how "that's communisim" (yes, I spelled that wrong intentionally)  while being blissfully ignorant of the reality of the game, as it currently stands. I can play a great game in a loosing battle and be rewarded less than I would be for a garbage game in a winning battle. Players are already being unfairly rewarded and penalized, every single battle of every single day. 



bockscar43 #13 Posted May 19 2019 - 19:26

    Captain

  • -Players-
  • 31501 battles
  • 1,570
  • [PL1AR] PL1AR
  • Member since:
    01-01-2015
 "You can't do it. It is impossible. This is why SBMM sucks, and should never be implemented in WoT."  Wot, will do whatever is necessary to increase its bottom line...just business not personal, people should remember that.                                                           

Guido1212 #14 Posted May 19 2019 - 19:32

    Community Contributor

  • Players
  • 80183 battles
  • 8,563
  • [CARTL] CARTL
  • Member since:
    06-11-2011

View PostDeviouslyCursed, on May 19 2019 - 16:59, said:

I already know my SBMM method would yield an eventual 50% win rate (using wins and losses to increase or decrease a rating that is used to place that person in a match). Some people have suggested other methods they think will lead to fair matches but not a 50% win rate. Personally, that doesn't make any sense, as keeping people who win more from continuing to win more is precisely the point of SBMM, but whatever. 

 

So, lets try a different approach. We'll use the a method of SBMM described in a previous post: Placing players of similar skill on opposite teams, balancing the teams. For instance, if 4 green players in a match, 2 green per side: 10 yellow, then 5 yellow per side, etc.

 

Now, some people claim that this won't screw over better players, because better players win more close games and that's how they have their high win rate. What is actually true is once a player has their high rating, they will win half of their 50% games, but they still win more games because due to random MM, they get more 51%+ chance to win games (it's because their high rating is being calculated into the chance to win). For some reason, it is difficult to convince people of this, and there is this continuous proclamation of: But good people win more battles when evenly matched, so therefore all the good players will still somehow keep winning more than 50%.

 

It's the equivalent of saying "Yeah, it's balanced, but then magic happens, and good people still win more, see?" Unfortunately, this can't happen. You see, if you have an equivalent player on the other side, then for each fight, there are two "players who are good and should win more" and one loses, the other wins (net of 50%). No matter how many battles you run, no matter how many are on each side, as long as it is balanced (same number on each side) some number of good players win, and an equal number of good players lose.

 

Match 1

Team1 vs Team 2

Team1: B G G G Y Y Y Y O O O R R R R

Team2: B G G G Y Y Y Y O O O R R R R

 

Team 1 wins

The result for wins/losses by skill gradient:

W/L

0/0

1/1

3/3

4/4

3/3

4/4

 

You can repeat that match as many times as you want. It won't matter. Regardless of who wins, the end result is the same. 50% win rate between the players of equal skill. This is true for 1 battle. It is true for 1k battles. What's worse, if someone from one of the skill brackets does manage to win more, they have pushed another player from that bracket BELOW 50%, which is completely insane if you think that is fair.

 

And BTW, it only gets worse when you consider that Draws are losses for both sides...

 

So, after 10 battles:

W/L

0/0

10/10

30/30

40/40

30/30

40/40

 

And as I already said, if they don't split those matches 50%, then 1 blue player is below 50%, 3 green players are below 50%, 4 red players are above 50%, etc.

 

And holy hell, it's doesn't matter if this is just "speculation." It is the application of logic, reason, and simple math.

 

I challenge anyone who thinks SBMM won't drive people to 50%, and/or screw over good players unjustly, to show teams and battle results that somehow show this! You have free reign to make team composition however you like and what ever team you want to win, as long as the teams are balanced. You can make different teams each match (still has to be balanced). Go ahead. Show me how this magic is supposed to work. Show me how you can place greens against greens, blues against blues, and somehow not end up with that category at 50% win rate (each player will either be 50% even, or some of those "good players who always manage to win more than they lose" will actual lose more than 50% so that another person in that category can win more than 50%).

 

Spoiler

 

 

I will always be amazed at those who believe equal results are "fair" and endeavor to create systems that force equal results.  There's nothing "fair" about being forced to a 50% W/R to create the illusion of fairness and "equal" matches.  In fact, it punishes the good players and helps the bad players.  Every time you excel, you get pushed to a worse team, and every time you fail you get pushed to a better team. 

 

That's not fair in any way shape or form. 

 

It's not even equal, nor does it make fair matches It simply attempts to apportion a W/R across a playerbase.  Bad players will still be bad, and good players will still be good.  Know how you'll know?  DPG, spotting, hit rate, and any number of other "stats".

 

Here's what is absolutely fair about this game.  Anyone can DL it and create an account.  That's perfectly fair.  From there it's up to you and all the myriad of factors that go into various stats to include W/R.  That's perfectly fair.  An even playing field at the first load in.  From there, it's all about you.

 

 

 



the_dude_76 #15 Posted May 19 2019 - 19:32

    Major

  • Players
  • 33377 battles
  • 4,542
  • [GSRM] GSRM
  • Member since:
    12-27-2011

Professional sports teams are all full of "purple" players so why isn't there parity in any pro sports league anywhere on the planet?? Shouldn't they all be winning and losing 50% of their games?? Obviously not...

 

oh and to the OP- You should probably look up the word "proving" in the dictionary because I don't think you understand what it means...



SporkBoy #16 Posted May 19 2019 - 20:04

    Major

  • Players
  • 44684 battles
  • 2,259
  • [PZB] PZB
  • Member since:
    02-06-2014
OP thinks he defined something but actually didn't. Still talking past each other and just blowing hot air

Hellsfog #17 Posted May 19 2019 - 20:09

    Major

  • Players
  • 37626 battles
  • 6,122
  • [VILIN] VILIN
  • Member since:
    06-22-2011

View Postthe_dude_76, on May 19 2019 - 13:32, said:

Professional sports teams are all full of "purple" players so why isn't there parity in any pro sports league anywhere on the planet?? Shouldn't they all be winning and losing 50% of their games?? Obviously not...

 

oh and to the OP- You should probably look up the word "proving" in the dictionary because I don't think you understand what it means...

 

5% of the server population is, statistically anyway, considered very good with 1% being considered great across all tiers and times of play. Grats on using an inapplicable analogy to illustrate a non-problem. 

 

View PostSporkBoy, on May 19 2019 - 14:04, said:

OP thinks he defined something but actually didn't. Still talking past each other and just blowing hot air

 

From your last three posts, you are trying way to hard to troll and doing it poorly. 


Edited by Hellsfog, May 19 2019 - 20:10.


TrevorsT112 #18 Posted May 19 2019 - 20:14

    Staff sergeant

  • -Players-
  • 9104 battles
  • 253
  • [MSFIT] MSFIT
  • Member since:
    10-31-2017

View PostDeviouslyCursed, on May 19 2019 - 17:59, said:

I already know my SBMM method would yield an eventual 50% win rate (using wins and losses to increase or decrease a rating that is used to place that person in a match). Some people have suggested other methods they think will lead to fair matches but not a 50% win rate. Personally, that doesn't make any sense, as keeping people who win more from continuing to win more is precisely the point of SBMM, but whatever. 

 

So, lets try a different approach. We'll use the a method of SBMM described in a previous post: Placing players of similar skill on opposite teams, balancing the teams. For instance, if 4 green players in a match, 2 green per side: 10 yellow, then 5 yellow per side, etc.

 

Now, some people claim that this won't screw over better players, because better players win more close games and that's how they have their high win rate. What is actually true is once a player has their high rating, they will win half of their 50% games, but they still win more games because due to random MM, they get more 51%+ chance to win games (it's because their high rating is being calculated into the chance to win). For some reason, it is difficult to convince people of this, and there is this continuous proclamation of: But good people win more battles when evenly matched, so therefore all the good players will still somehow keep winning more than 50%.

 

It's the equivalent of saying "Yeah, it's balanced, but then magic happens, and good people still win more, see?" Unfortunately, this can't happen. You see, if you have an equivalent player on the other side, then for each fight, there are two "players who are good and should win more" and one loses, the other wins (net of 50%). No matter how many battles you run, no matter how many are on each side, as long as it is balanced (same number on each side) some number of good players win, and an equal number of good players lose.

 

Match 1

Team1 vs Team 2

Team1: B G G G Y Y Y Y O O O R R R R

Team2: B G G G Y Y Y Y O O O R R R R

 

Team 1 wins

The result for wins/losses by skill gradient:

W/L

0/0

1/1

3/3

4/4

3/3

4/4

 

You can repeat that match as many times as you want. It won't matter. Regardless of who wins, the end result is the same. 50% win rate between the players of equal skill. This is true for 1 battle. It is true for 1k battles. What's worse, if someone from one of the skill brackets does manage to win more, they have pushed another player from that bracket BELOW 50%, which is completely insane if you think that is fair.

 

And BTW, it only gets worse when you consider that Draws are losses for both sides...

 

So, after 10 battles:

W/L

0/0

10/10

30/30

40/40

30/30

40/40

 

And as I already said, if they don't split those matches 50%, then 1 blue player is below 50%, 3 green players are below 50%, 4 red players are above 50%, etc.

 

And holy hell, it's doesn't matter if this is just "speculation." It is the application of logic, reason, and simple math.

 

I challenge anyone who thinks SBMM won't drive people to 50%, and/or screw over good players unjustly, to show teams and battle results that somehow show this! You have free reign to make team composition however you like and what ever team you want to win, as long as the teams are balanced. You can make different teams each match (still has to be balanced). Go ahead. Show me how this magic is supposed to work. Show me how you can place greens against greens, blues against blues, and somehow not end up with that category at 50% win rate (each player will either be 50% even, or some of those "good players who always manage to win more than they lose" will actual lose more than 50% so that another person in that category can win more than 50%).

 

Spoiler

 

 



SporkBoy #19 Posted May 19 2019 - 20:18

    Major

  • Players
  • 44684 battles
  • 2,259
  • [PZB] PZB
  • Member since:
    02-06-2014
Claims simple math but offers not one equation, relation, function for review.

SporkBoy #20 Posted May 19 2019 - 20:20

    Major

  • Players
  • 44684 battles
  • 2,259
  • [PZB] PZB
  • Member since:
    02-06-2014
Where is the definition or 'balance' or 'fair' - not trolling to point about fatal flaws and poor logic




1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users