Jump to content


Matchmaking Discussion Thread

MM

  • Please log in to reply
11409 replies to this topic

Long_Range_Sniper #11201 Posted 11 January 2019 - 06:40 PM

    General

  • Player
  • 34097 battles
  • 9,258
  • [-MM] -MM
  • Member since:
    04-04-2011

View Postm1x_angelico, on 11 January 2019 - 12:45 AM, said:

Well, few pages ago we had a "git gut" discussion, so I posted data that shows that git gut advice doesnt really make sense in terms of current MM, or to be more precise it brings very limited returns, at best. It is an enlightened philosophy in general, but not applicable here. If a player can hope to achieve a 58% winrate instead of 50% if he/she improves to an excellent skill level, that means that out of 1000 matches, he will win 580 instead of 500 (80 matches difference where his supreme skills influenced the outcome), while the remaining 420 matches remain outside of his control, so to speak (this is not actually correct, because collapsing team issue is distributed among wins and losses, but shows that getting gud does not get you a very high corresponding win rate).

 

Dunno why moderator edited the post, as this was clearly not naming and shaming. If that was naming and shaming, then the XVM tool itself is naming and shaming on a much grander scale,  because I can see stats of all players in the match. Off course if I want to share some data with forumites, they must be able to verify it, otherwise I can photoshop the picture to oblivion.

 

Actually this is not the correct way to understand the "influence" you have in a game with statistical floors and ceilings.

 

It is not the case that a player winning 58% wins 80 matches due to their influence from 1000, and 420 are outside of their control as it is outlined above. This is because there are floors and ceilings that the player will not go below or above if playing as part of the general population.

 

If you take 45% and 65% as the floor and ceiling for winrate performance, then the actual workable area to influence as a player is the space in between.

 

From 1000 games this is the space between 450 games and 650 games. ie....200 games.

 

So a 45% player is the baseline and they win no more games at all. (zero out of 200)

A 50% player wins 50 games more (winning 500 games) - 50/200

A 58% player is winning 130 more (winning 580 games) - 130/200

 

Which means that out of 200 "winnable" games a 58% player wins 65% and a 50% player wins 25%.

 

65% compared to 25% shows the real influence a 58% winrate player has on the potential outcome of a game in comparison to a 50% player. It's almost three times the influence.

 

Which is why you get the XVM percentiles shaped the way they do.

 

Image result for xvm winrate percentiles

 

So no, 420 games are not outside of your control, because 400 games are outside of everyone's control. 

 

Edit: Which has been explained by a few players as well in the thread.

 

View Postm1x_angelico, on 11 January 2019 - 12:45 AM, said:

Well, few pages ago we had a "git gut" discussion, so I posted data that shows that git gut advice doesnt really make sense in terms of current MM,

 

So as explained above, if you understand the real difference between a 58% player and a 50% player, "git gud" in this MM is the best advice.


Edited by Long_Range_Sniper, 11 January 2019 - 06:44 PM.


m1x_angelico #11202 Posted 11 January 2019 - 07:23 PM

    Second Lieutenant

  • Player
  • 23668 battles
  • 1,045
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View PostSoup_Potato, on 11 January 2019 - 05:40 PM, said:

Is this still with your definition of "collapsing teams" which you haven't actually defined other than "10 players dying suddenly". That isn't realistic really is it? Like i said ages ago it seems that your dedinition of "collapsing" might easily be 2 or 3 players dying from one team before any die from the other. Still no example replay or anything. And no, i wont be clicking your own created download links. Upload the replay(s) to wotreplays where its safe and easy to download them from. Why not run Baldrickk's analyser as well with those replays you have?

 

Dont really understand, who's definition am I supposed to use in my analysis?  You got the replays, so download them and do your analysis as you see fit, with the definitions you see fit. Dont want to download them? Then dont. I really cant be expected to deal with your paranoia issues.

 

I'm continuing the discussion with data (although small sample size atm), while you are continuing with the same baseless nonsense...

 

View Posteekeeboo, on 11 January 2019 - 06:20 PM, said:

Moderation is easier on the forums because I can check everything and read back as can mods. People when they report things can be found. 

In-battle people will generally not report people for either shared view or because they've run out of reports, reporting arty instead etc. 

The forum name and shame, for those who were around at the time, are to prevent the dark times of people witch hunting on the forums. 

 

I understand the need for it. Still over-moderation is the other side of the coin as under-moderation. If I write on the forum the same data that is publicly and freely available to everyone, how is this naming and shaming, while public display of this same data by WG itself is not? That's what I mean when I say moderation here is sometimes weird. But, it's not a biggy.



eekeeboo #11203 Posted 11 January 2019 - 07:29 PM

    English Community Manager

  • WG Staff
  • 45838 battles
  • 929
  • Member since:
    07-25-2010

View Postm1x_angelico, on 11 January 2019 - 06:23 PM, said:

 

Dont really understand, who's definition am I supposed to use in my analysis?  You got the replays, so download them and do your analysis as you see fit, with the definitions you see fit. Dont want to download them? Then dont. I really cant be expected to deal with your paranoia issues.

 

I'm continuing the discussion with data (although small sample size atm), while you are continuing with the same baseless nonsense...

 

 

I understand the need for it. Still over-moderation is the other side of the coin as under-moderation. If I write on the forum the same data that is publicly and freely available to everyone, how is this naming and shaming, while public display of this same data by WG itself is not? That's what I mean when I say moderation here is sometimes weird. But, it's not a biggy.

 

Because that data isn't available anywhere, not to mention data that's inaccurate like people who started witch hunts like "This player is a bot" - because that player stole their kill and thus you with hunt a ban. That behaviour and attitude has no place in this game. I'm all for no naming and shaming, because as much as it's an inconvenience sometimes. It's better to not end up with an innocent person unjustly harassed and abused because someone decided they want to make life difficult for someone. 

m1x_angelico #11204 Posted 11 January 2019 - 07:35 PM

    Second Lieutenant

  • Player
  • 23668 battles
  • 1,045
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View PostLong_Range_Sniper, on 11 January 2019 - 06:40 PM, said:

 

Actually this is not the correct way to understand the "influence" you have in a game with statistical floors and ceilings.

 

It is not the case that a player winning 58% wins 80 matches due to their influence from 1000, and 420 are outside of their control as it is outlined above. This is because there are floors and ceilings that the player will not go below or above if playing as part of the general population.

 

If you take 45% and 65% as the floor and ceiling for winrate performance, then the actual workable area to influence as a player is the space in between.

 

From 1000 games this is the space between 450 games and 650 games. ie....200 games.

 

So a 45% player is the baseline and they win no more games at all. (zero out of 200)

A 50% player wins 50 games more (winning 500 games) - 50/200

A 58% player is winning 130 more (winning 580 games) - 130/200

 

Which means that out of 200 "winnable" games a 58% player wins 65% and a 50% player wins 25%.

 

65% compared to 25% shows the real influence a 58% winrate player has on the potential outcome of a game in comparison to a 50% player. It's almost three times the influence.

 

Which is why you get the XVM percentiles shaped the way they do.

 

Image result for xvm winrate percentiles

 

So no, 420 games are not outside of your control, because 400 games are outside of everyone's control. 

 

Edit: Which has been explained by a few players as well in the thread.

 

 

So as explained above, if you understand the real difference between a 58% player and a 50% player, "git gud" in this MM is the best advice.

 

Ok, let us hypothetically say that 45% and 65% are ceilings (I say hypothetically, because I haven't seen data to support this).

 

This means that compared to an average player having 50% winrate, if you get gut to such a superb, godlike level, you will get +15% winrate. This means that even if this would be possible for most of the players to get to this level of skill, compared to an average player, in 150 (out of 1000) matches that would have been otherwise lost, as a godlike player, you will turn them to victory. 

 

What about the other matches, besides these 15% of matches where you managed to influence the match due to your godlike skill?



Baldrickk #11205 Posted 11 January 2019 - 07:42 PM

    Field Marshal

  • Player
  • 31151 battles
  • 15,069
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postm1x_angelico, on 11 January 2019 - 07:35 PM, said:

 

Ok, let us hypothetically say that 45% and 65% are ceilings (I say hypothetically, because I haven't seen data to support this).

 

This means that compared to an average player having 50% winrate, if you get gut to such a superb, godlike level, you will get +15% winrate. This means that even if this would be possible for most of the players to get to this level of skill, compared to an average player, in 150 (out of 1000) matches that would have been otherwise lost, as a godlike player, you will turn them to victory. 

 

What about the other matches, besides these 15% of matches where you managed to influence the match due to your godlike skill?

Those battles are still influenced, but not enough to cause a win.

 

If we take Muffin's dataset for example, he has more 15:0 wins than losses, because he's good enough to turn what might have been a 15:3 into a 15:0 or a 0:15 into a 3:15.  Not always, but some of the time.

 

As Long range says, you should look at WR in comparison to the attained range of results to see how influence on the win rate scales.

 


Edited by Baldrickk, 11 January 2019 - 07:43 PM.


m1x_angelico #11206 Posted 11 January 2019 - 07:43 PM

    Second Lieutenant

  • Player
  • 23668 battles
  • 1,045
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View Posteekeeboo, on 11 January 2019 - 07:29 PM, said:

Because that data isn't available anywhere, not to mention data that's inaccurate like people who started witch hunts like "This player is a bot" - because that player stole their kill and thus you with hunt a ban. That behaviour and attitude has no place in this game. I'm all for no naming and shaming, because as much as it's an inconvenience sometimes. It's better to not end up with an innocent person unjustly harassed and abused because someone decided they want to make life difficult for someone. 

 

I understand you, but I dont think we are talking about the same thing. The stats I mentioned are freely and publicly available and provided by WG on their homepage: LINK

If we are discussing with data as basis, then we must name the source or provide data, otherwise, we can pull numbers and stats from are respective bottoms.

18:47 Added after 4 minutes

View PostBaldrickk, on 11 January 2019 - 07:42 PM, said:

Those battles are still influenced, but not enough to cause a win.

If we take Muffin's dataset for example, he has more 15:0 wins than losses, because he's good enough to turn what might have been a 15:3 into a 15:0 or a 0:15 into a 3:15.  Not always, but some of the time.

As Long range says, you should look at WR in comparison to the attained range of results to see how influence on the win rate scales.

 

So there is a significant % of matches where you can influence the match, but not the final outcome in terms of whether it is a win or defeat?



jabster #11207 Posted 11 January 2019 - 07:48 PM

    Field Marshal

  • Beta Tester
  • 12614 battles
  • 24,464
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View Postm1x_angelico, on 11 January 2019 - 06:23 PM, said:

 

Dont really understand, who's definition am I supposed to use in my analysis?  You got the replays, so download them and do your analysis as you see fit, with the definitions you see fit. Dont want to download them? Then dont. I really cant be expected to deal with your paranoia issues.

 

I'm continuing the discussion with data (although small sample size atm), while you are continuing with the same baseless nonsense...

 

 

I understand the need for it. Still over-moderation is the other side of the coin as under-moderation. If I write on the forum the same data that is publicly and freely available to everyone, how is this naming and shaming, while public display of this same data by WG itself is not? That's what I mean when I say moderation here is sometimes weird. But, it's not a biggy.

 

Holding up some specifc player for possible ridicule is hardly the same thing is it?

m1x_angelico #11208 Posted 11 January 2019 - 07:58 PM

    Second Lieutenant

  • Player
  • 23668 battles
  • 1,045
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View Postjabster, on 11 January 2019 - 07:48 PM, said:

Holding up some specifc player for possible ridicule is hardly the same thing is it?

 

That's what moderation is for. You moderate the people ridiculing. How am I supposed to discuss based on data, if I'm not allowed to state the data (which the other party can verify), which was provided by the WG itself?



Long_Range_Sniper #11209 Posted 11 January 2019 - 08:04 PM

    General

  • Player
  • 34097 battles
  • 9,258
  • [-MM] -MM
  • Member since:
    04-04-2011

View Postm1x_angelico, on 11 January 2019 - 06:43 PM, said:

So there is a significant % of matches where you can influence the match, but not the final outcome in terms of whether it is a win or defeat?

 

If you're influencing a significant amount of games significantly then you're significantly increasing the chances it will be a win. As long as you accept that for 33% of the time no matter what you do, it will be a loss no matter how you play.

 

A 60% player is way way more influential than a 50% player as it's not a 10% difference between them in terms of performance against their peers.

 

"Git gud" is the way to be influential as the maths shows.

 

View Postm1x_angelico, on 11 January 2019 - 06:35 PM, said:

What about the other matches, besides these 15% of matches where you managed to influence the match due to your godlike skill?

 

It's not 15% of matches through godlike skill.

 

58% is 580 games from that 1000 game sample, and 50% is 500 games. which gives 80 games difference from 200 possible "winnable" games and is 40% difference.

 

You're influencing 40% more than a 50% player if you're a 58% player and that's over and above a player who's already at the 63% percentile.

 

The "godlike" results come from looking at where you sit on those percentiles.


Edited by Long_Range_Sniper, 11 January 2019 - 08:09 PM.


jabster #11210 Posted 11 January 2019 - 08:11 PM

    Field Marshal

  • Beta Tester
  • 12614 battles
  • 24,464
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View Postm1x_angelico, on 11 January 2019 - 06:58 PM, said:

 

That's what moderation is for. You moderate the people ridiculing. How am I supposed to discuss based on data, if I'm not allowed to state the data (which the other party can verify), which was provided by the WG itself?

 

Post it without the name and if anyone really want's to verify it then you can PM them the name. Not that hard is it surely?

Baldrickk #11211 Posted 11 January 2019 - 08:15 PM

    Field Marshal

  • Player
  • 31151 battles
  • 15,069
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postm1x_angelico, on 11 January 2019 - 07:43 PM, said:

So there is a significant % of matches where you can influence the match, but not the final outcome in terms of whether it is a win or defeat?

Well yeah. 

You play in every single one of your battles, you have some impact,  for better or for worse in every single one. 



DangerMouse #11212 Posted 11 January 2019 - 08:24 PM

    Warrant Officer

  • Beta Tester
  • 27567 battles
  • 796
  • [FLOG] FLOG
  • Member since:
    10-28-2010

View Posteekeeboo, on 11 January 2019 - 05:20 PM, said:

Moderation is easier on the forums because I can check everything and read back as can mods. People when they report things can be found. 

 

In-battle people will generally not report people for either shared view or because they've run out of reports, reporting arty instead etc. 

 

The forum name and shame, for those who were around at the time, are to prevent the dark times of people witch hunting on the forums. 

 

I always like a nice witch hunt, that ducking stool is really going to waste at the moment.

 

 

Trying to get from 50% to 54% WR is tough, getting close but having a few ups and downs, every % greater than 50 gets harder and harder, your influence has to be so much more for that very reason that you have a limited scope of influence. Its the close games that you have to turn in to wins.

 

DM


Edited by DangerMouse, 11 January 2019 - 08:28 PM.


m1x_angelico #11213 Posted 12 January 2019 - 07:38 AM

    Second Lieutenant

  • Player
  • 23668 battles
  • 1,045
  • [-VETO] -VETO
  • Member since:
    01-04-2015

View Postjabster, on 11 January 2019 - 08:11 PM, said:

Post it without the name and if anyone really want's to verify it then you can PM them the name. Not that hard is it surely?

 

Not hard, but highly inefficient.

 

View PostBaldrickk, on 11 January 2019 - 08:15 PM, said:

Well yeah. 

You play in every single one of your battles, you have some impact,  for better or for worse in every single one. 

 

That's my point Baldrickk. If it finally shows that there is 40% of collapsing team matches (let's say distributed evenly - 50% wins/losses), that means that you get 400 matches out of 1000 where your skill doesnt matter in a way that would prevent the team from collapsing or in a way that your team would lose if you didnt play. In these matches you can help get a better result in terms of you WN8 and numbers of opposing players killed, but that's it. In 40% matches where skill is a factor, but you will get even number of wins/losses, in matches which are not collapsing teams, and in 20% matches you will have the opportunity where the level of your skill may directly influence the outcome. That means that you only have to actually play in 60% of matches to get your current winrate, opposed to playing in 100% matches.

 

As a result, if you are a 65% winrate player, concerned only about the winrate, the issues for you may be: 1)those 50% losses from the collapsing team and 2)inability to directly influence more than 20% of matches. However, it can be also argued that since you will get 50% losses/wins with collapsing team matches, that you dont actually care, because in those matches you can relax as they will not impact your winrate.

 

If you are a 65% winrate player,who doesnt really care about the winrate, but about games being fun, the issues for you are: 1) 40% collapsing team matches 

 

This is why players who dont care about winrate are 2x more invested in fixing of the MM. If we take into account that winrate player doesnt even care about 50% losses from collapsing teams, that we see that those players who dont care about winrate are even more interested in fixing of the MM.

 


Edited by m1x_angelico, 12 January 2019 - 07:40 AM.


LordMuffin #11214 Posted 12 January 2019 - 09:27 AM

    Field Marshal

  • Player
  • 48529 battles
  • 11,459
  • [-GLO-] -GLO-
  • Member since:
    06-21-2011

View Postm1x_angelico, on 12 January 2019 - 07:38 AM, said:

 

Not hard, but highly inefficient.

 

 

That's my point Baldrickk. If it finally shows that there is 40% of collapsing team matches (let's say distributed evenly - 50% wins/losses), that means that you get 400 matches out of 1000 where your skill doesnt matter in a way that would prevent the team from collapsing or in a way that your team would lose if you didnt play. In these matches you can help get a better result in terms of you WN8 and numbers of opposing players killed, but that's it. In 40% matches where skill is a factor, but you will get even number of wins/losses, in matches which are not collapsing teams, and in 20% matches you will have the opportunity where the level of your skill may directly influence the outcome. That means that you only have to actually play in 60% of matches to get your current winrate, opposed to playing in 100% matches.

 

As a result, if you are a 65% winrate player, concerned only about the winrate, the issues for you may be: 1)those 50% losses from the collapsing team and 2)inability to directly influence more than 20% of matches. However, it can be also argued that since you will get 50% losses/wins with collapsing team matches, that you dont actually care, because in those matches you can relax as they will not impact your winrate.

 

If you are a 65% winrate player,who doesnt really care about the winrate, but about games being fun, the issues for you are: 1) 40% collapsing team matches 

 

This is why players who dont care about winrate are 2x more invested in fixing of the MM. If we take into account that winrate player doesnt even care about 50% losses from collapsing teams, that we see that those players who dont care about winrate are even more interested in fixing of the MM.

 

Nah.

~40% of the games being won no matter what you do doesn't imply that ~40% of your games being lost whatever you do.

 

Also, you don't know which ~40% you always win, only that over a large number of battles, roughly 40% will be won no matter what.

So in order to achieve a good winrate, you are forced to play as good as you can every battle and try to maximize your effect on the outcome every time.


Edited by LordMuffin, 12 January 2019 - 09:34 AM.


jabster #11215 Posted 12 January 2019 - 09:39 AM

    Field Marshal

  • Beta Tester
  • 12614 battles
  • 24,464
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View Postm1x_angelico, on 11 January 2019 - 04:27 PM, said:

Don't want to go into discussing moderation, which can be weird at times, but it is clear why I posted those stats and it certainly wasn't to shame the said player (having low skill is nothing to be ashamed of).

 

I agree there are 14 others players in the team, and this is exactly what I 've been talking about all the time. If your role in the team is limited by the said team, then getting good has a very limited influence at best, as shown. Again, certain players were trying to explain that other 14 players dont matter as long as you git gut. These stats point to another conclusion.

 

If you are the kind of player that is focused on winrates and nothing else matters in this game, is +8% higher winrate, or even 10% worth all the crappy matches with the collapsing teams that the rest of us (who dont care about +8% winrate) have to suffer?

 

FYI summary from 71 random matches played in 2019 (45% matches with collapsed teams, out of those 31% wins/68%losses):

 

Spoiler

 

 

In case I’ve missed it what defintion are you using for a collapsing team as it’s pointless trying to dicuss a problem if you don’t even know what the problem is.

Baldrickk #11216 Posted 12 January 2019 - 10:10 AM

    Field Marshal

  • Player
  • 31151 battles
  • 15,069
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postjabster, on 12 January 2019 - 09:39 AM, said:

 

In case I’ve missed it what defintion are you using for a collapsing team as it’s pointless trying to dicuss a problem if you don’t even know what the problem is.

If its the same as I've seen before, it'll include 

15:5 games. 

Which is a problem. 

 

Analysis of results puts 15:5 to 15:7 as the most common outcomes, all roughly equal, with results either side dropping off, and fairly heavily as we approach 15:0.

 

This corresponds with some rough experiments I've previously put together to model how scores should be, which predicted that for teams the size they are, with permadeath, the snowball effect is a very strong force - it pushes the expected result out to 15:5-6.  I've talked about this before. 

 

This doesn't give much scope for real collapsing teams :( I'd suggest 15:3 to be the limit, but whether that should be inclusive or not? I don't know.

The numbers definitely show that those results are actually really rare.


Edited by Baldrickk, 12 January 2019 - 10:11 AM.


jabster #11217 Posted 12 January 2019 - 10:20 AM

    Field Marshal

  • Beta Tester
  • 12614 battles
  • 24,464
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostBaldrickk, on 12 January 2019 - 09:10 AM, said:

If its the same as I've seen before, it'll include 

15:5 games. 

Which is a problem. 

 

Analysis of results puts 15:5 to 15:7 as the most common outcomes, all roughly equal, with results either side dropping off, and fairly heavily as we approach 15:0.

 

This corresponds with some rough experiments I've previously put together to model how scores should be, which predicted that for teams the size they are, with permadeath, the snowball effect is a very strong force - it pushes the expected result out to 15:5-6.  I've talked about this before. 

 

This doesn't give much scope for real collapsing teams :( I'd suggest 15:3 to be the limit, but whether that should be inclusive or not? I don't know.

The numbers definitely show that those results are actually really rare.

 

Would model did you use for results as I did look at this very briefly, several years ago, and the snowball factor I tried to use just meant the vast majority of games ended as landslides.

 

As for the term collapsing terms, it still seems too vague to use in anything but general terms.


Edited by jabster, 12 January 2019 - 10:38 AM.


Baldrickk #11218 Posted 12 January 2019 - 10:41 AM

    Field Marshal

  • Player
  • 31151 battles
  • 15,069
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postjabster, on 12 January 2019 - 10:20 AM, said:

 

Would model did you use for results as I did look at this very briefly, several years ago, and the snowball factor I tried to use just meant the vast majority of games ended as landslides.

 

As for the term collapsing terms, it still seems to vague to use in anything but general terms.

Agreed, there is no accepted definition,  which is why I support your request for his terms.

 

As for the model, it was fairly rough, but good enough to demonstrate the point:

  • Two teams, each "player" with randomly assigned "skill" level.   For the purposes of this they all start with the same HP 
  • Players on each team split up between "flanks" 1,2,3 and "campers/arty", again, randomly. 
  • Each "tick" of the simulation,  tanks on flanks 1-3 attempt to damage an enemy on the same flank. Success in dealing damage determined by their relative skill value vs RNG.  Penalties applied for overcrowded flanks (lemming trains).  "Campers" can damage any flank,  but are heavily penalised.
  • Once a flank is won, winning tanks all move together to a randomly chosen location where enemy tanks are and continue fighting

 

It misses a lot,  but it was only meant to roughly emulate permadeath teams, not be a perfect reflection of WOT



jabster #11219 Posted 12 January 2019 - 10:51 AM

    Field Marshal

  • Beta Tester
  • 12614 battles
  • 24,464
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostBaldrickk, on 12 January 2019 - 09:41 AM, said:

Agreed, there is no accepted definition,  which is why I support your request for his terms.

 

As for the model, it was fairly rough, but good enough to demonstrate the point:

  • Two teams, each "player" with randomly assigned "skill" level.   For the purposes of this they all start with the same HP 
  • Players on each team split up between "flanks" 1,2,3 and "campers/arty", again, randomly. 
  • Each "tick" of the simulation,  tanks on flanks 1-3 attempt to damage an enemy on the same flank. Success in dealing damage determined by their relative skill value vs RNG.  Penalties applied for overcrowded flanks (lemming trains).  "Campers" can damage any flank,  but are heavily penalised.
  • Once a flank is won, winning tanks all move together to a randomly chosen location where enemy tanks are and continue fighting

 

It misses a lot,  but it was only meant to roughly emulate permadeath teams, not be a perfect reflection of WOT

 

Interesting. When I looked at it I was trying something far more abstract with using a team’s overall skill as a bias to who loses a tank next. Seeing the results of the model I thought am I interested enough to try and create something more complicated. I also had this concern that even if I produced a model that more closely matched real results had I convinced myself it wasn’t rigged or instead that the model I used matched what a rigged system produces. That’s when I got back to work wondering what is it about buffer overflows people don’t understand while telling me it’s fine as “it will never happen”.

Baldrickk #11220 Posted 12 January 2019 - 10:55 AM

    Field Marshal

  • Player
  • 31151 battles
  • 15,069
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postjabster, on 12 January 2019 - 10:51 AM, said:

 

Interesting. When I looked at it I was trying something far more abstract with using a team’s overall skill as a bias to who loses a tank next. Seeing the results of the model I thought am I interested enough to try and create something more complicated. I also had this concern that even if I produced a model that more closely matched real results had I convinced myself it wasn’t rigged or instead that the model I used matched what a rigged system produces. That’s when I got back to work wondering what is it about buffer overflows people don’t understand while telling me it’s fine as “it will never happen”.

What do you mean there is only a limited amount of bu







Also tagged with MM

2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users