Jump to content


How team balance (win chance) affects your battles

XVM SBMM blowout roflstomp

  • Please log in to reply
221 replies to this topic

NeatoMan #61 Posted Jun 05 2019 - 12:25

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostSimplyPzB2, on Jun 05 2019 - 01:46, said:

You say I'm wrong, yet didn't 'fix' the data.  Someone knows he's wrong...

another fine example of how budha only sees what he wants to see.  The answer was right above his last post, yet he conveniently ignored it, because it wasn't what he wanted to hear.

 

here is what that particular set of data is telling us:

When the underdog team beats a heavily favored team it usually doesn't do it by blowing out the better team...     that's all



spud_tuber #62 Posted Jun 05 2019 - 17:06

    Major

  • Players
  • 59355 battles
  • 8,878
  • Member since:
    08-26-2013

View PostSimplyPzB2, on Jun 05 2019 - 00:46, said:

 

You say I'm wrong, yet didn't 'fix' the data.  Someone knows he's wrong...

 

Stop expecting others to do your work for you.   Fix it yourself now that you know it is wrong. 



WeSayNotToday #63 Posted Jun 05 2019 - 20:00

    Captain

  • -Players-
  • 23953 battles
  • 1,182
  • Member since:
    04-08-2015

View Postspud_tuber, on Jun 03 2019 - 16:14, said:

Folks,if you can't find a way to have fun in steamrolls, you're probably playing the wrong game.

 

Maybe, but more likely, you're probably in a non-meta tank, or you got caught out of position, or you don't like farming while losing.

 

Some might then say being in a non-meta tank is the problem, of the game-within-the-game.... :sceptic:

 

 



ez_money #64 Posted Jun 05 2019 - 20:40

    Captain

  • Players
  • 15818 battles
  • 1,651
  • Member since:
    10-23-2011

View PostNeatoMan, on Jun 02 2019 - 14:43, said:

 

Here are two, a 3005 PR vs 6696 PR player

 

You can also see how much your PR shifts the curve below.  Though it doesn't account for tier, so if you have a 6k PR and play low tiers it will have a greater impact than if you play tier 10,  The setup below is for mainly a tier 7-10 population

https://docs.google....dit?usp=sharing

 

View PostCopacetic, on Jun 02 2019 - 14:46, said:

seems sbmm would have a minimum effect if any

 

I disagree. Just based on the sample graph above you are looking at a difference of about 55% WR versus 43% WR between the two players based on PR. Although we know PR is nebulous at best, it still highlights the point. A team of predominantly 3000 PR players will almost always get stomped by the team of mostly 6000 PR players. If the MM is truly random, then you would expect to find yourself on either of those type of teams roughly half the time, over a long enough period of time. I.E. you will be on the team getting stomped as much as being on the team doing the stomping pretty much equally.

 

But...That assumes that the composition of teams and the matchups of teams is truly random, and it would have to be over a large enough number of battles to make it non-anomalous. What number of battles would that need to be? 100? 10,000? 100,000? I personally don't know, but we can assume it will be a substantial number. On a day-to-day basis people are playing a much lower number of battles at a time. Given this low number of battles, you are much more susceptible to a "losing streak" or a "winning streak" on any given day(s) which would be outside the win/lose probabilities previously discussed (potentially way outside). E.G. I am about 54% WR/7300 PR, and I went 2-15 yesterday and 25-35 over the past 7 days. That means I have been on teams with more of the 3000 PR players than the ones with the 6000 PR players (I use XVM, so I can attest that has indeed been the case lately).


 

SBMM would help close the gap and reduce the "streakiness" of the game by putting roughly equal teams against each other...the 3000's playing each other and the 6000's each other separately and/or an equal balance of both on each team (the latter would be preferable, otherwise the 3000's will likely never get any better). The 300's will hopefully get better, and the 6000's will have to play up to their ratings (or better) against an equally matched opposing team, and everybody can rise with the tide.



NeatoMan #65 Posted Jun 05 2019 - 22:45

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View Postez_money, on Jun 05 2019 - 14:40, said:

SBMM would help close the gap and reduce the "streakiness" of the game by putting roughly equal teams against each other...the 3000's playing each other and the 6000's each other separately and/or an equal balance of both on each team (the latter would be preferable, otherwise the 3000's will likely never get any better). The 300's will hopefully get better, and the 6000's will have to play up to their ratings (or better) against an equally matched opposing team, and everybody can rise with the tide.

It won't change the streaky results.  From all the data I've gathered most streaks include some favored teams and balanced teams that simply blew it.  SBMM won't change that

 

Block Quote

A team of predominantly 3000 PR players will almost always get stomped by the team of mostly 6000 PR players.

and that setup rarely happens.

 



SimplyPzB2 #66 Posted Jun 05 2019 - 22:54

    First lieutenant

  • -Players-
  • 105 battles
  • 647
  • Member since:
    05-26-2016

View Postez_money, on Jun 05 2019 - 20:40, said:

 

 

I disagree. Just based on the sample graph above you are looking at a difference of about 55% WR versus 43% WR between the two players based on PR. Although we know PR is nebulous at best, it still highlights the point. A team of predominantly 3000 PR players will almost always get stomped by the team of mostly 6000 PR players. If the MM is truly random, then you would expect to find yourself on either of those type of teams roughly half the time, over a long enough period of time. I.E. you will be on the team getting stomped as much as being on the team doing the stomping pretty much equally.

 

But...That assumes that the composition of teams and the matchups of teams is truly random, and it would have to be over a large enough number of battles to make it non-anomalous. What number of battles would that need to be? 100? 10,000? 100,000? I personally don't know, but we can assume it will be a substantial number. On a day-to-day basis people are playing a much lower number of battles at a time. Given this low number of battles, you are much more susceptible to a "losing streak" or a "winning streak" on any given day(s) which would be outside the win/lose probabilities previously discussed (potentially way outside). E.G. I am about 54% WR/7300 PR, and I went 2-15 yesterday and 25-35 over the past 7 days. That means I have been on teams with more of the 3000 PR players than the ones with the 6000 PR players (I use XVM, so I can attest that has indeed been the case lately).


 

SBMM would help close the gap and reduce the "streakiness" of the game by putting roughly equal teams against each other...the 3000's playing each other and the 6000's each other separately and/or an equal balance of both on each team (the latter would be preferable, otherwise the 3000's will likely never get any better). The 300's will hopefully get better, and the 6000's will have to play up to their ratings (or better) against an equally matched opposing team, and everybody can rise with the tide.

 

Well put.  It's almost as if it's a super simple concept.  (It is).  No idea why the 'random  mm fanboys' refuse to get this....

ExploratorOne #67 Posted Jun 05 2019 - 23:19

    Staff sergeant

  • -Players-
  • 15812 battles
  • 371
  • [LOU1] LOU1
  • Member since:
    12-31-2015

One major thing that this analysis does not factor in is whether or not the players are playing to win.  With all of the various missions in play at any given time, each player may be playing for something other than to win.  For instance, I spent a week or two before the last Frontline grinding towards a couple of Tier 8 tanks I wanted; the same thing I did up until a week ago.  That forced me to play some vehicles I did not like or do well in.  Currently, the D-Day missions are putting some players in less preferred tanks.  Then there are the Campaigns.  In all cases, players may be going for damage or kills or XP.

 

DISCLAIMER:  I was relieved when win chance disappeared.  It caused too many players to change their play style. i.e., give up before we started.


Edited by ExploratorOne, Jun 05 2019 - 23:20.


NeatoMan #68 Posted Jun 05 2019 - 23:30

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostExploratorOne, on Jun 05 2019 - 17:19, said:

One major thing that this analysis does not factor in is whether or not the players are playing to win.  With all of the various missions in play at any given time, each player may be playing for something other than to win.  For instance, I spent a week or two before the last Frontline grinding towards a couple of Tier 8 tanks I wanted; the same thing I did up until a week ago.  That forced me to play some vehicles I did not like or do well in.  Currently, the D-Day missions are putting some players in less preferred tanks.  Then there are the Campaigns.  In all cases, players may be going for damage or kills or XP.

 

DISCLAIMER:  I was relieved when win chance disappeared.  It caused too many players to change their play style. i.e., give up before we started.

 

which is why more data sets will clear that up.  As I added more and more data from different players I found the numbers CONVERGED, rather than diverged, which indicates that whatever is going on is pretty much standard behavior. 

 

I did get a set of data from someone who has platooned a bit recently, and not surprisingly he overperformed the win chances, but it did maintain a straight line fit, just offset a bit, which is another good indication.



NeatoMan #69 Posted Jun 06 2019 - 01:15

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostWeSayNotToday, on Jun 05 2019 - 14:00, said:

Some might then say being in a non-meta tank is the problem, of the game-within-the-game.... :sceptic:

agreed.  most of my tanks have fallen out of meta, and many of my tier 10s have been left to rot for years without any buffs, while more and more OP ones in other trees keep getting released.  I still play them, but my win rate suffers.   Their tank balancing department is drunk



SimplyPzB2 #70 Posted Jun 06 2019 - 05:09

    First lieutenant

  • -Players-
  • 105 battles
  • 647
  • Member since:
    05-26-2016

View PostNeatoMan, on Jun 05 2019 - 23:30, said:

 

which is why more data sets will clear that up.  As I added more and more data from different players I found the numbers CONVERGED, rather than diverged, which indicates that whatever is going on is pretty much standard behavior. 

 

I did get a set of data from someone who has platooned a bit recently, and not surprisingly he overperformed the win chances, but it did maintain a straight line fit, just offset a bit, which is another good indication.

 

Just one example of being able to beat predicted results...  if one good player toons, and one good player doesn't. The tooner will have a higher win rate.

-

And playing non-mata tanks, another fine example of being able to beat predicted results..  if one good player plays only meta, and the other plays a variety, meta man will have a higher win rate...



Bezzell #71 Posted Jun 06 2019 - 06:49

    Sergeant

  • Beta Testers
  • 31144 battles
  • 247
  • [MRE] MRE
  • Member since:
    07-26-2010
Waiting for the tick-box to make my stats anonymous.  Blocking XVM will solve a lot of problems. 

NeatoMan #72 Posted Jun 06 2019 - 12:33

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostSimplyPzB2, on Jun 05 2019 - 23:09, said:

Just one example of being able to beat predicted results...  if one good player toons, and one good player doesn't. The tooner will have a higher win rate.

-

And playing non-mata tanks, another fine example of being able to beat predicted results..  if one good player plays only meta, and the other plays a variety, meta man will have a higher win rate...

Yep,  it's as I said all along, with skill minimized in SBMM, bringing OP tanks, platooning, and RNG become more of a determining factor

 

Block Quote

Well put.  It's almost as if it's a super simple concept.  (It is).  No idea why the 'random  mm fanboys' refuse to get this....

 When your only method of analysis involves counting your fingers, and then making up whatever you want, of course everything seems super simple.

 



NeatoMan #73 Posted Jun 06 2019 - 21:47

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

I hear a number of people say extreme win chance games are guaranteed wins and losses.    Well, not quite.  Here is individual game WN8s for our test group, broken down by win chance.  You can see the win rates they achieved for a certain level of performance in each win chance group

 

 

For the most part whenever they had a decent game (>1200 WN8) they outperformed the win chances.  Even in the lowest WC group they could double their chances by having a good game.  Only at the highest end does it seem the contribution, or lack of it, had little effect on the outcome.  I'll have to look to see how many of those very high win chances were generated by platoons.  That might explain the lack of WR movement vs individual contribution.   In most other ranges the win rates were swayed by 30% or more depending on the individual contribution. 

 

As with all the other analyses, this could use more data.  Some of the sample sizes are small, but the trends already seem to be in place.

 

bottom line is your contributions matter, so try to win even when the odds are against you.


Edited by NeatoMan, Jun 06 2019 - 23:45.


Jer1413 #74 Posted Jun 06 2019 - 23:16

    Captain

  • Players
  • 48220 battles
  • 1,557
  • [RR13] RR13
  • Member since:
    02-24-2013

View PostNeatoMan, on Jun 06 2019 - 16:47, said:

I hear a number of people say extreme win chance games are guaranteed wins and losses.    Well, not quite.  Here is individual game WN8s for our test group, broken down by win chance.  You can see the win rates they achieved for a certain level of performance in each win chance group

 

 

For the most part whenever they had a decent game (>1200 WN8) they outperformed the win chances.  Even in the lowest WC group they could double their chances by having a good game.  Only at the highest end does it seem the contribution, or lack of it, had little effect on the outcome.  I'll have to look to see how many of those very high win chances were generated by platoons.  That might explain the lack of WR movement vs individual contribution.   In most other ranges the win rates were swayed by 30% or more depending on the individual contribution. 

 

As with all the other analyses, this could use more data.  Some of the sample sizes are small, but the trends already seem to be in place.

 

bottom line is your contributions matter, so try to win even when the odds are against you.

 

Imagine that, play well, even in an "unbalanced" poor win-chance "guaranteed loss", and you can still win your fair share.

 


Edited by Jer1413, Jun 06 2019 - 23:16.


NeatoMan #75 Posted Jun 06 2019 - 23:56

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostJer1413, on Jun 06 2019 - 17:16, said:

Imagine that, play well, even in an "unbalanced" poor win-chance "guaranteed loss", and you can still win your fair share.

You'll still win near what the win chances indicate, but if you can manage to do well, all is not lost.  You never know which one of those will go your way, so might as well not get discouraged.  I consider those games expert mode.



SimplyPzB2 #76 Posted Jun 07 2019 - 00:39

    First lieutenant

  • -Players-
  • 105 battles
  • 647
  • Member since:
    05-26-2016

View PostNeatoMan, on Jun 06 2019 - 21:47, said:

I hear a number of people say extreme win chance games are guaranteed wins and losses.    Well, not quite.  Here is individual game WN8s for our test group, broken down by win chance.  You can see the win rates they achieved for a certain level of performance in each win chance group

 

 

For the most part whenever they had a decent game (>1200 WN8) they outperformed the win chances.  Even in the lowest WC group they could double their chances by having a good game.  Only at the highest end does it seem the contribution, or lack of it, had little effect on the outcome.  I'll have to look to see how many of those very high win chances were generated by platoons.  That might explain the lack of WR movement vs individual contribution.   In most other ranges the win rates were swayed by 30% or more depending on the individual contribution. 

 

As with all the other analyses, this could use more data.  Some of the sample sizes are small, but the trends already seem to be in place.

 

bottom line is your contributions matter, so try to win even when the odds are against you.

 

Neato's data proving Neato wrong since 2011.  Neato swears you 'can't beat the statistics'.   Yet, here he posts examples of people doing just that.

-

Skill balanced mm would have the effect of just making 40-50% ctw and 50-60% ctw battles.  So to see what Sbmm would look like, and what peoples win rate would be under sbmm, all we have to do is just look at the results from these two brakets:

                     40-50       50-60     Avg  (note, there were 1,023 40-50 and 1,243 50-60 - but I've just added and divided by two for the 'avg' wr/)

0-600             33%         39%       36%      (New/Casual players with low skill NOT pulled up to 50%)

600-1200       44%         56%       50%      (Average players get an average win rate)

1200-2000     43%         57%       50%      (Good players get an average win rate)

2000-3000     59%         67%       63%      (Very good player do NOT get pulled down to 50%)

3000+            67%         77%       72%      (Great players do NOT get pulled down to 50%)

-

LOOK AT THOSE RESULTS!!!!  According you Neato and the random mm fanboys - ALL PLAYER WILL BE FORCED TO 50% W/R UNDER SBMM.   That's the holy of holy statements.  Yet, really bad players stay well under 50%, as the 0-600 only average 36%.  THEY ARE NOT PULLED UP!   Average players (600-1200 and 1200-2000) amazingly have average results at 50%. Very good and great players have win rates of 63% and 72% - THEY ARE NOT PULLED DOWN!.   Amazing.  For the umpteeth thousandth time, the data clearly shows better player can and will do better than 50% under sbmm, and bad player will still do less than 50%. 

-

Thanks Neato, I will be using this data/post a lot.  It really clears up the issue....

-

(Also, I agree, never quit just because you are on a team that is going to lose.  You can still farm kills, damage, and xp...)

 



NeatoMan #77 Posted Jun 07 2019 - 02:24

    Major

  • Players
  • 28209 battles
  • 20,779
  • Member since:
    06-28-2011

View PostSimplyPzB2, on Jun 06 2019 - 18:39, said:

"Look at me!  I just completely butchered some more numbers again!"

YOU ARE SO WRONG!!!   Don't ever try to math logic.  You always get it completely wrong

 

 

Block Quote

                     40-50       50-60     Avg  (note, there were 1,023 40-50 and 1,243 50-60 - but I've just added and divided by two for the 'avg' wr/)

0-600             33%         39%       36%      (New/Casual players with low skill NOT pulled up to 50%)

600-1200       44%         56%       50%      (Average players get an average win rate)

1200-2000     43%         57%       50%      (Good players get an average win rate)

2000-3000     59%         67%       63%      (Very good player do NOT get pulled down to 50%)

3000+            67%         77%       72%      (Great players do NOT get pulled down to 50%)

 NO NO NO NO NO!!!

 

That's not at all what those numbers represent.  You are so off base I don't even know where to start to set you straight (if that's at all possible).  Maybe try reading my post one word at a time.  Take a real long time to let every word soak in.  Maybe then you'll realize the complete lunacy you just posted.

 

or maybe go through that table one item at a time and tell me what you think each item represents, so I can get an idea what's going through that tiny little ball of crap in your head that you call a brain.

 

It would help if you took the time to actually figure out what exactly those numbers mean before opening your mouth, and if something isn't clear then just ask.  

...but nope, that's not the budha way... "see numbers... make something up with those numbers"


Edited by NeatoMan, Jun 07 2019 - 03:31.


DeviouslyCursed #78 Posted Jun 07 2019 - 05:00

    First lieutenant

  • -Players-
  • 4949 battles
  • 740
  • Member since:
    12-13-2018

Oh man. This thread reminds me of why I changed majors from Biomedical Engineering after 3 years despite having a 3.6 GPA (at a University that was ranked #2 in the nation in BME, btw).

 

I have to give it to Neatoman, he knows his statistics. I would think long and hard before disagreeing with him on anything statistics related.



DeviouslyCursed #79 Posted Jun 07 2019 - 05:02

    First lieutenant

  • -Players-
  • 4949 battles
  • 740
  • Member since:
    12-13-2018

View PostNeatoMan, on Jun 04 2019 - 20:58, said:

My first reaction if I saw that data would be "something is wrong with your formula".   But unlike you I know where it's coming from and the context. 

 

I refuse to discuss anything dealing with numbers or statistics with you any more.  You have absolutely no idea how to handle them.  You are completely incapable of having a logical discussion when it comes to math.  You are on a completely different world

 

From now on you will get the following response from me every single time you post your nonsense

 

I would suggest just posting the following

 

View PostDeviouslyCursed, on Jun 02 2019 - 05:11, said:

View PostSimplyPzB2, on Jun 02 2019 - 02:43, said:

 

Ask neato to show you the data for the 10-19% ctw and 80-89% ctw brackets he has.  What you will see is that the team with 10%+ ctw (for example) will win less than 10% of these battles.  And you will see the team with 80%+ will win more than 90% of these battles.  So the point is, getting put into 'the type' of battles where you will likely win 2-3 out of every hundred of these types of battles IS A BAD THING.  And getting put into 'the type' of battles where you will only lose 2-3 out of 100 IS A BORING THING.  And while Neato will scream 'these don't happen often', they do.  Certainly for a game where many players have played, 10k, 30k, 80k battles.  Do the math, if these are 1% of your battles and you've played 30k battles at 6min average battle time - you've wasted 60 hours of your life (half bad, half boring).  Thing is, these happen to varying degrees, about 20% of all your battles.  Do the math again 60 hours x 19 = 1,140 hours of wasted gameplay.   That's why this matters.

 

1% of 30k is 300

(300 battles x 6 min) / 60 (mins in hour) = 30 hours

How did you get 60 hours?

 

Then, if you wanted to find out how many hours that would be for 20% instead of 1%, you would multiple that by 20, not 19. It's actually 600 hours, btw. I figured I'd save you the embarrassment of getting it wrong... again.

 

You can't even do simple math yet you think you know how SBMM will work, and how to interpret data and statistics. You are such a tool.

 

It was his own post, his own example, and his own complete and total failure at basic math.


Edited by DeviouslyCursed, Jun 07 2019 - 05:04.


DeviouslyCursed #80 Posted Jun 07 2019 - 05:09

    First lieutenant

  • -Players-
  • 4949 battles
  • 740
  • Member since:
    12-13-2018

View PostNeatoMan, on Jun 04 2019 - 16:09, said:

View PostgrandthefttankV, on Jun 03 2019 - 00:29, said:

OP you need a hobby

Actually, I think I got too many...   not enough time for all of them.

 

The sign of intelligence is that one is not bored when given a large amount of free time. I've seen so often people try to retire and then go back to work because they "couldn't stand having nothing to do." I've never understood this. If I never had to work again for the rest of my life, I would still have something to do every day till the day I died.

 

It doesn't surprise me at all that Neatoman is in the same category.







Also tagged with XVM, SBMM, blowout, roflstomp

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users