Jump to content


Official Matchmaking Discussion Thread


  • Please log in to reply
1513 replies to this topic

j_rod #1421 Posted 08 August 2020 - 10:33 PM

    Not a moderator, but Better Looking Than Cletus

  • Players
  • 28050 battles
  • 2,384
  • [III-R]
  • Member since:
    05-04-2011
CA, you’re an *** and yes, I understand statistics and probability. When I ask for “data” what I’m asking for is a statistical breakdown that shows that MM isn’t random over a large data set. If you need me to give you an exact explanation of that that looks like, I’ll happily explain it in greater detail once I can type from a computer and not a phone.

Short of showing that, nothing you say is anything more than anecdotes.

Working as intended
 

It was never fair... but it was fun! - Krietenstein34 

 


j_rod #1422 Posted 08 August 2020 - 10:50 PM

    Not a moderator, but Better Looking Than Cletus

  • Players
  • 28050 battles
  • 2,384
  • [III-R]
  • Member since:
    05-04-2011

View PostCA_vampire, on 08 August 2020 - 02:55 AM, said:

One more thing: For a computer programmer is easy to write the code for the matchmaker, even if it uses a complicated strategy comparing past wins, a hidden time-to-lose variable etc etc etc.

 

It is just a comparison of numbers, for less than 1,000 tanks. This is fast and easy.  I mean it is easy when comparing it to other programming aspects of this game, like the 3D models, animations, smooth and realistic movement, collisions, etc. The MM is one of the easiest modules.

 

And it is natural that they have programmer(s) for that, and it is natural for the programmers to experiment with various strategies, especially since this is something hidden on the server. They can be changing the MM often and none of us would realize it. Obviously, they decided to keep one of their experiments, hence the Ratings. 

 

Of course, the goal of the MM is to make you addicted to the game. That's the main goal of every game out there! 

 

 


Nobody is saying that it isn’t plausible that MM is rigged. That’s what you fail to understand. We all think it’s possible but short of seeing the actual code and algorithm, we have to make conclusions off the “data” that we can see, which you continue to try to minimize. 


Working as intended
 

It was never fair... but it was fun! - Krietenstein34 

 


_Crusader6_ #1423 Posted 09 August 2020 - 05:45 PM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014

View PostTexas_Tyrant, on 08 August 2020 - 07:09 AM, said:

IF MM is rigged explain the correlation between damage out put and winning. I do agree with you though that it is very easy to rig, I believe though that it is fair as possible. Its random as I can see. Besides that there is no proof of correlation between paying for goodies and WR.

 

Yeehaw,

Texas_Tyrant


Exactly. 
  We can all see the ‘average player’ stats on Blitzstars- and those are players who’ve been looked up.   
 One still finds a lot of players who aren’t, and those are general low 40% players.  
 

While I don’t doubt CA that it’s possible to rig - the why is my biggest issue. 
 A near random MM system is going to throw more Odd results that lead to the ‘random draw’ that causes addiction in this, without major effort or any liability on WG’s end. 
   They’ve repeatedly told us a lot of what isn’t in MM, and those are generally the things we’d expect to see. 
1) No vehicle weighting beyond type, expect for new vehicle honeymoon where it tries to put you top tier for the first 8-15 games depending on queue size etc.    I’ve had new vehicles low tier, so clearly time to battle is emphasized more than the honeymoon.  
more on this in #4

2) no overall W/L or tank W/L or recent W/L counted. 
3) maps are counted, so it tries not to repeated load you in the same map.  
 

There is a lot they don’t say about MM. 

 And frankly it’s always seem odd when you get streaky sessions as low tier (which isn’t just observation bias, as I’ve logged a lot of games and see a lot of low tier in certain tanks.) 

 

4) As well the whole aspect of long weights in some vehicles.  WG has admittedly that certain rare vehicles are tried to be pitted against each other - which brings to question vehicle weighting comments in #1, unless they don’t consider that vehicle weighting...

 

But at the end of the day in a 500 series battle session - generally nothing looks out of the ordinary, when looking at all the open source data for green and red teams.  
 


 

 


 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 


CA_vampire #1424 Posted 10 August 2020 - 06:51 AM

    __Banned__

  • Players
  • 30288 battles
  • 1,238
  • [MOMB]
  • Member since:
    11-21-2016

View Post_Crusader6_, on 03 August 2020 - 06:55 AM, said:


Or one can simply look at the player base, and realize that [....]

  

 


 

 

What you are implying, in multiple posts, is that the majority of the players are useless morons. Well, I agree with you, but that's not the point here. 

 

My focus is not the players, it is the WG programmers. These programmers are smarter than the average guy out there (not just the average player!). These programmers (I don't know how many they have) go to work every day, and they get paid good money for writing code. Go read the reviews on Google Play store. How many reviews mention the MM+RNG? A lot. So, do you think that WG has zero programmers working on MM+RNG? But if they do have programmers working on MM+RNG, what do you think that these people are getting paid for? What do they do day in and day out? You seem to believe that they just have a completely random MM+RNG that works basically the same for many years now, and nothing has changed in the last 2-3 years (for example) and nothing needs to change. You really seem to believe they don't tinker with it at all. Do you really find this plausible? If you do, then -based on my experience as a programmer- I am telling you that you are wrong.

 

That's what I said in my initial post (and the following posts, too). What I am saying is nothing complicated and I am not degrading you or some other posters that seem to get offended for no reason at all!

 

(If anyone is interested, go back and read again my posts in the last 2-3 pages and you will see for yourself that I am not saying anything outrageous here.  ;)

 

 

 

 


No Mercy / No Malice

_Crusader6_ #1425 Posted 10 August 2020 - 02:00 PM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014

CA, I’m not dismissing what you say.   
  We’ve seen WG has tweaked RNG for pen and DMG several times.  
As well we’ve seen several MM changes. 
 

What I’m saying is in the MM changes WG has worked with it in Open Test.   
   Generally for those changes they have disclosed what they are doing.   Now are they doing something more?  Perhaps, and probably. 
 
But if anything is occurring, it is fairly certain it’s not a large issue - or we wouldn’t see 80% players or 30% players.  
  
If you remember Face4star, he cane up with EWR when he went on tilt - doe he was convinced that WG was rigging based on data he was seeing on WOTBSTARS.    
  I’m a relatively smart guy for a knuckledragger - but I’m accepting I know zilch about code and what can or can’t be done.  
What I am saying is by and large I believe whatever WG does isn’t causing any undue results over time.  I believe that simply from looking at the data sets - and while I get frustrated at times when RNG gives me the thumbs down - I see it happen the other way at times, and have faith it all balances out in the end.  

Interestingly enough reading back on some of your comments, I’ve noticed that WG was silent about if RNG pays attention to W/L results and while they stated that MM doesn’t - now they said RNG is calculated at the then of the shot - I don’t know if there is some underlying modifications done to RNG based on previous results.   
 

Last comment. 
   I’ve been low tier 48 of my last 56 games.  
 It makes one start to look for boogeymen in the closet at time. 
 Fortunately I know if I find a boogeyman, he’s going to be the one scared ;) 

 

 


 

 


 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 


jtp59 #1426 Posted 10 August 2020 - 06:10 PM

    Private

  • Players
  • 7386 battles
  • 5
  • Member since:
    07-17-2019
I think the rng is more rigged than the matchmaker. I started getting pretty mouthy in battles and my accuracy went to crap. I mean like missing the first 4 or 5 shots of every battle.

_Crusader6_ #1427 Posted 10 August 2020 - 07:44 PM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014

View Postjtp59, on 10 August 2020 - 01:10 PM, said:

I think the rng is more rigged than the matchmaker. I started getting pretty mouthy in battles and my accuracy went to crap. I mean like missing the first 4 or 5 shots of every battle.


Confirmation bias ;) 

 

Mind you 2,800 avg dmg in my IS-8 today was good for 1 win 3 loss.   The win was a terribad 1,300 dmg game where I got racked early.  Go figure. 
   

   
 


 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 


Sniper_Hanz #1428 Posted 10 August 2020 - 10:15 PM

    Lance-corporal

  • Players
  • 7322 battles
  • 18
  • [RT166]
  • Member since:
    03-09-2016

View PostCA_vampire, on 10 August 2020 - 06:51 AM, said:

 

What you are implying, in multiple posts, is that the majority of the players are useless morons. Well, I agree with you, but that's not the point here. 

 

My focus is not the players, it is the WG programmers. These programmers are smarter than the average guy out there (not just the average player!). These programmers (I don't know how many they have) go to work every day, and they get paid good money for writing code. Go read the reviews on Google Play store. How many reviews mention the MM+RNG? A lot. So, do you think that WG has zero programmers working on MM+RNG? But if they do have programmers working on MM+RNG, what do you think that these people are getting paid for? What do they do day in and day out? You seem to believe that they just have a completely random MM+RNG that works basically the same for many years now, and nothing has changed in the last 2-3 years (for example) and nothing needs to change. You really seem to believe they don't tinker with it at all. Do you really find this plausible? If you do, then -based on my experience as a programmer- I am telling you that you are wrong.

 

That's what I said in my initial post (and the following posts, too). What I am saying is nothing complicated and I am not degrading you or some other posters that seem to get offended for no reason at all!

 

(If anyone is interested, go back and read again my posts in the last 2-3 pages and you will see for yourself that I am not saying anything outrageous here.  ;)

 

 

 

 

 

Of all the comments I've read so far about MM (WR and RNG, included) yours deserve my congrats. Double thumbs up.



CA_vampire #1429 Posted 11 August 2020 - 04:13 AM

    __Banned__

  • Players
  • 30288 battles
  • 1,238
  • [MOMB]
  • Member since:
    11-21-2016

A little something for those of you who like ... data! 

 

Assume I am a WG programmer. I want to compare two scenarios: 

 

A. Random RNG.

B. Slightly helpful RNG. If a player loses 4 battles in a row, the player gets 20% more damage per shot and receives 20% less damage per shot. 

 

I arrange the server software to change during reboots in the following way: August 1st: scenario A, August 2nd: scenario B, August 3rd: scenario A, August 4th: scenario B and so on. I collect data for three months and compute if the players on the average played more battles per day under scenario A or under scenario B or there is no difference. I make some plots and give a presentation to my manager. Obviously, as a company we want our players to play the game a lot. 

 

(This is very easy to implement for a programmer. And of course, if this is your job and you work on these things for 8 hours per day for 5+ years... you can invent much more complicated scenarios. And yes, some of these things do make a difference and the software companies do these things all the time.  )

 

======

 

Now assume I am a random player. I collect ... "data". I don't know what to measure. I have no idea if the server software changes, how often they change it, and what changes they make. I have no meaningful boundaries for the "data". What is my "data" worth?    (Answer: It is not worth a fart!  ) 

 


Edited by CA_vampire, 11 August 2020 - 04:15 AM.

No Mercy / No Malice

j_rod #1430 Posted 11 August 2020 - 01:01 PM

    Not a moderator, but Better Looking Than Cletus

  • Players
  • 28050 battles
  • 2,384
  • [III-R]
  • Member since:
    05-04-2011

View PostCA_vampire, on 10 August 2020 - 10:13 PM, said:

A little something for those of you who like ... data! 

 

Assume I am a WG programmer. I want to compare two scenarios: 

 

A. Random RNG.

B. Slightly helpful RNG. If a player loses 4 battles in a row, the player gets 20% more damage per shot and receives 20% less damage per shot. 

 

I arrange the server software to change during reboots in the following way: August 1st: scenario A, August 2nd: scenario B, August 3rd: scenario A, August 4th: scenario B and so on. I collect data for three months and compute if the players on the average played more battles per day under scenario A or under scenario B or there is no difference. I make some plots and give a presentation to my manager. Obviously, as a company we want our players to play the game a lot. 

 

(This is very easy to implement for a programmer. And of course, if this is your job and you work on these things for 8 hours per day for 5+ years... you can invent much more complicated scenarios. And yes, some of these things do make a difference and the software companies do these things all the time.  )

 

======

 

Now assume I am a random player. I collect ... "data". I don't know what to measure. I have no idea if the server software changes, how often they change it, and what changes they make. I have no meaningful boundaries for the "data". What is my "data" worth?    (Answer: It is not worth a fart!  ) 

 


CA, I’m 100% with you that any “data” on RNG is difficult to collect but it is possible if one should want to. All you’d have to do is track the games played on WOTinspector and pop the damage per shot in to an Excel spreadsheet. That said, what you’re suggesting is very plausible.

 

However, this thread is about MM, not RNG. 


Working as intended
 

It was never fair... but it was fun! - Krietenstein34 

 


_Crusader6_ #1431 Posted 12 August 2020 - 12:23 PM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014

WG has claimed they don’t do that for RNG on pen/dmg.  
  Also kills monkey with avg dmg / shot - so you need to remove those when you’re tracking that.  
 

IMHO your better off thinking about dispersion/accuracy if you are looking for an easy place to monkey.   
   I’m not sure WG has even given their distribution method for dispersion - there have been several ideas floated - but if your trying to influence a player’s performance.  
 

That said I still don’t see the why.   The mystery of team composition is part of the draw to the game.  
   Given we see some 80% players and some high 20% players - if WG is putting in some help/hinder features - they aren’t designed to do much.   
 


 


 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 


__V_O_P__ #1432 Posted 16 August 2020 - 12:23 AM

    First Sergeant

  • Players
  • 48070 battles
  • 1,279
  • [LAP]
  • Member since:
    09-06-2016
CA the main bone of contention is not that MM isn’t random - because it isn’t -
It’s a group of people who feel singled out by WG for something (not spending money, swearing, being too good, being too bad, not being a member of the society) and the way that they think WG punishes them is through MM giving them worse teams.

That MM code may have changed is irrelevant since people have complained over long periods of time. That we don’t see the MM code is also irrelevant because these people cannot offer anything beyond their own personal observations that they have been personally selected. I’m interested in seeing any evidence of favoritism or a lack of it - like anything. Do I have standards for evidence? Yes I guess but they’d not be very demanding.



__V_O_P__ #1433 Posted 16 August 2020 - 12:42 AM

    First Sergeant

  • Players
  • 48070 battles
  • 1,279
  • [LAP]
  • Member since:
    09-06-2016
On an entirely different note...

One thing WG could change is the medium-light balance of power in MM. I played a couple of games recently when one team had one lower tier med-light and the other team had two (one high and one lower) med-lights. Seemed to me that anytime there’s only one tank on a flank which is out gunned and out numbered that’s not good but when it’s the med-light flank that just seems to be something that significantly predicts battle outcome.

Thoughts? Should MM allow for one flank to be out numbered and out gunned - I understand that it obviously there’s balance here (on the heavy flank) but imo the med-light flank is a strong predictor of battle outcome (just my opinion at the moment).

_Crusader6_ #1434 Posted 16 August 2020 - 02:30 AM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014
IMHO balance by tank type sucks. 
  It makes for cookie cutter repetition, and it still won’t satisfy some who will still be clamoring for more, like balance of each tank to be mirrored on red. Then When that doesn’t satisfy 7 v 7 of the exactly same tank.  
 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 


__V_O_P__ #1435 Posted 16 August 2020 - 01:10 PM

    First Sergeant

  • Players
  • 48070 battles
  • 1,279
  • [LAP]
  • Member since:
    09-06-2016

View Post_Crusader6_, on 16 August 2020 - 02:30 AM, said:

IMHO balance by tank type sucks. 
  It makes for cookie cutter repetition, and it still won’t satisfy some who will still be clamoring for more, like balance of each tank to be mirrored on red. Then When that doesn’t satisfy 7 v 7 of the exactly same tank.  


agreed. I don’t want exactly the same tanks either but whenever I see one lower tier med on my team and red have two I think “this is going to smart”



whatzup22 #1436 Posted 31 August 2020 - 06:16 AM

    Sergeant

  • Players
  • 3038 battles
  • 423
  • [FOIL]
  • Member since:
    03-25-2017

View PostCA_vampire, on 11 August 2020 - 12:13 PM, said:

A little something for those of you who like ... data! 

 

Assume I am a WG programmer. I want to compare two scenarios: 

 

A. Random RNG.

B. Slightly helpful RNG. If a player loses 4 battles in a row, the player gets 20% more damage per shot and receives 20% less damage per shot. 

 

I arrange the server software to change during reboots in the following way: August 1st: scenario A, August 2nd: scenario B, August 3rd: scenario A, August 4th: scenario B and so on. I collect data for three months and compute if the players on the average played more battles per day under scenario A or under scenario B or there is no difference. I make some plots and give a presentation to my manager. Obviously, as a company we want our players to play the game a lot. 

 

(This is very easy to implement for a programmer. And of course, if this is your job and you work on these things for 8 hours per day for 5+ years... you can invent much more complicated scenarios. And yes, some of these things do make a difference and the software companies do these things all the time.  )

 

======

 

Now assume I am a random player. I collect ... "data". I don't know what to measure. I have no idea if the server software changes, how often they change it, and what changes they make. I have no meaningful boundaries for the "data". What is my "data" worth?    (Answer: It is not worth a fart!  ) 

 

err, you think WG programmers bother to write all this code out? I dont think so, they should be more focused on buffs and nerfs, new tank stats, program, etc.



ODDEO_GUY #1437 Posted 05 September 2020 - 03:33 PM

    Lance-corporal

  • Players
  • 10438 battles
  • 77
  • [T_Q_T]
  • Member since:
    08-30-2019

Figured I'd put this here, as it''s basically just another rant regarding the "unfairness" or at least "uneven-ness" of the random MM system.

 

My T-34-85 is ALWAYS bottom tier.

It's a tier 6, an "even" tier that supposedly will see top tier games more frequently than odd tiers.

 

I've played 69 games in it.

6 were top tier.

 

Discuss, if you wish.

Ignore if you desire.

Just saying. It gets real old at some point.

Yes, I understand it doesn't accomplish anything to talk about it.

I now leave this commentary to sell everything related to it, ammo, consumables, provisions, and park it.

Sucks cause I spent a lot of free XP on modules, wasted.

But whatever....



CA_vampire #1438 Posted 06 September 2020 - 01:13 AM

    __Banned__

  • Players
  • 30288 battles
  • 1,238
  • [MOMB]
  • Member since:
    11-21-2016

I had a mission win 3 in a row. Two easy wins but the third was a bad loss.

 

Has this happened to you? (Of course it has!!!)

 

Do you really think this is completely random? Do you think that the programmers made no effort to fix or manipulate this? Do you really believe the programmers left it completely random? Why would you think so? They are programmers, they can do whatever they like. They are the real "gods" in this game (or any game). 

 

Here is my team in that third game: 

 

And here is what they were like, from top to bottom ... either new players or 40% "veterans":

 

 


Edited by CA_vampire, 06 September 2020 - 01:16 AM.

No Mercy / No Malice

CA_vampire #1439 Posted 06 September 2020 - 01:27 AM

    __Banned__

  • Players
  • 30288 battles
  • 1,238
  • [MOMB]
  • Member since:
    11-21-2016

View Postwhatzup22, on 30 August 2020 - 11:16 PM, said:

err, you think WG programmers bother to write all this code out? I dont think so, they should be more focused on buffs and nerfs, new tank stats, program, etc.

 

 

Bother? It is their job. And "randomness" is not automatic, you know? It has to be programmed, too. 

 

I have said it many times that a lot of people out there confuse "Computer Programs" with "Nature". In Nature, in many instances you have an intrinsic randomness, and if you want something structured you need extra effort. But Computer Programs are not like that! Not at all! EVERYTHING has to be programmed. EVERYTHING! Even "randomness" has to be programmed. And there are many cases that it is easier to program something deterministic than to program randomness. There is no reason to say that it is easier for some computer program to be "random" than to be "fixed". 

 

Computer Programs are NOT like Nature. At least please learn this simple fact. 

 

 

 

 

 

 

 

No Mercy / No Malice

_Crusader6_ #1440 Posted 06 September 2020 - 02:27 AM

    Vote Slime Mold

  • Players
  • 70713 battles
  • 15,692
  • [BDGER]
  • Member since:
    12-08-2014

Carry harder.

 

I am here to tell you, that WG does not need to do anything to have disastrous runs.

   I will upload a replay tomorrow that took an easy win to a draw because a muppet went town and took 3 other morons with him.

WG can't account for individual choice - and so many idiots abound.


Edited by _Crusader6_, 06 September 2020 - 02:29 AM.

 

 
Tank Hoarder: 424 tanks in Garage:  435/435 Played Tanks Aced
 
This space for rent until next terrible mechanic ;) 
 
    Wallet Warrior: Loyal Original M60 owner
 
 

 





1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users