Jump to content


matchmaking in 6.0 is absurd

matchmaking

  • Please log in to reply
24 replies to this topic

opavder_13 #1 Posted 07 June 2019 - 04:19 AM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015

Has anyone else observed that how frequent it is to get matchmaking where the average winrate of one team is ~60% and 40% of the other. On raising a ticket the same stupid reply comes from wg that they match tank by type not players. 


5 years is too long to still use this retarded algo for matchmaking.
Even if you are matching by type of tank, just flip the players for each tank type so as to balance out the average winrate. e.g 
Team A's awr is 58% with a 65% e50 and Team B awr is 44% with a 41% t54, then switch the e50 to team B and t54 to team A (awr - avg win rate)


I am getting too many game where my hts and tds are all moron with hts camping and rhm going to hull down in the front.



FOIL_Looney #2 Posted 07 June 2019 - 04:34 AM

    Sergeant

  • Member
  • 16341 battles
  • 201
  • [FOIL]
  • Member since:
    09-18-2015
I've had to wait in a queue for over 10 minutes before in a KV1S with over 20 tanks all the time at same, below and above tiers, so 60 tanks available, plus more than 30 tanks of each type available at all times . Clearly nothing to do with matchmaking by just tank type. Ticket with screenshots received no reply at all.

opavder_13 #3 Posted 07 June 2019 - 05:34 AM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015

Its so frustrating at times to loose games after games because my team is full of players with like 2k battles and 40% wr while enemy team has  65% plat with rest 50+ wr as well. So many of my friends have left this game because of this retreaded matchmaking and i am thinking the same. 

 

They spend so much effort on useless avatars but not on matchmaking without which it will be just another stupid multiplayer. 

 



Gandalf__Greyhame #4 Posted 07 June 2019 - 05:52 AM

    First Lieutenant

  • Member
  • 224 battles
  • 967
  • [SHLRE]
  • Member since:
    06-24-2016

View Postopavder_13, on 07 June 2019 - 04:19 AM, said:

Has anyone else observed that how frequent it is to get matchmaking where the average winrate of one team is ~60% and 40% of the other. On raising a ticket the same stupid reply comes from wg that they match tank by type not players. 


5 years is too long to still use this retarded algo for matchmaking.
Even if you are matching by type of tank, just flip the players for each tank type so as to balance out the average winrate. e.g 
Team A's awr is 58% with a 65% e50 and Team B awr is 44% with a 41% t54, then switch the e50 to team B and t54 to team A (awr - avg win rate)


I am getting too many game where my hts and tds are all moron with hts camping and rhm going to hull down in the front.

 

Show the evidence . You might get the odd game where there is a significant difference between wr but usually wr is within 2 or 3% . When there is a blue player on opposite team there will almost invariably be a blue player on your team as well . 

                                                            

 

                          

                 

   


Rinskiev #5 Posted 07 June 2019 - 06:03 AM

    Sergeant

  • Member
  • 10881 battles
  • 185
  • [AUSIE]
  • Member since:
    07-01-2014

I agree with Gandalf. People remember the cases with odd things happening than the usual case. Maybe you didn't understand the answer by WG and they should understand that not everybody knows about statistics: the select by tank type -> players' WR are random -> you may be lucky at times while at others unlucky. Did you check the WR of both teams when you win? OR...is this why I always win?



TnkDztroyer #6 Posted 07 June 2019 - 10:27 AM

    Sergeant

  • Member
  • 7 battles
  • 237
  • Member since:
    02-04-2019

View PostGandalf__Greyhame, on 07 June 2019 - 03:52 PM, said:

 

Show the evidence . You might get the odd game where there is a significant difference between wr but usually wr is within 2 or 3% . When there is a blue player on opposite team there will almost invariably be a blue player on your team as well . 

 

Nonsense!
When the KV-2 get nerfted, but the Smasher doesn't, we know something is rotten in Denmark. Can anyone say "Pay to Win"?

pegasoos #7 Posted 07 June 2019 - 11:59 AM

    Corporal

  • Member
  • 1092 battles
  • 14
  • [MAGU_]
  • Member since:
    11-09-2017

View PostTnkDztroyer, on 07 June 2019 - 10:27 AM, said:

 

Nonsense!

 

Ash has always been WG loyal to the bone! I do not know if either he owns a bunch of shares from WG or if he is some kind of [edited] fellow by principle. If he was a good player or had decent stats at least, it would make sense to defending his performance and being able to outplay MM. Or maybe because of being an underperforming player anyway, rigged MM does not make any difference for him?

​MM is very different on various accounts - you surely have noticed that part. The odds of having such loosing strikes (3-4 losses are already statistically odd when you start playing), clearly indicates rigging to incite players to play more, at least 10 games to recover their WR, or for those who do not care about WR, to have a few wins in a session. If you haven't played the game on a certain account for some time, you often get 1-2 Wins for starters, to motivate you to go on! MM is rigged - and that's that!

 

 
 


Edited by pegasoos, 07 June 2019 - 12:00 PM.


TnkDztroyer #8 Posted 07 June 2019 - 12:09 PM

    Sergeant

  • Member
  • 7 battles
  • 237
  • Member since:
    02-04-2019
Yep! the things that happen in this game can't be explained without there being manipulation of some kind
When the KV-2 get nerfted, but the Smasher doesn't, we know something is rotten in Denmark. Can anyone say "Pay to Win"?

TinDingo #9 Posted 08 June 2019 - 04:44 AM

    Sergeant

  • Member
  • 27243 battles
  • 230
  • [OVR50]
  • Member since:
    02-13-2016

View PostGandalf__Greyhame, on 07 June 2019 - 05:52 AM, said:

 

Show the evidence . You might get the odd game where there is a significant difference between wr but usually wr is within 2 or 3% . When there is a blue player on opposite team there will almost invariably be a blue player on your team as well . 

You and I are not playing the same game!


Every system is perfectly designed to produce exactly the results that it produces


opavder_13 #10 Posted 08 June 2019 - 12:40 PM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015

View PostGandalf__Greyhame, on 07 June 2019 - 05:52 AM, said:

 

Show the evidence . You might get the odd game where there is a significant difference between wr but usually wr is within 2 or 3% . When there is a blue player on opposite team there will almost invariably be a blue player on your team as well . 

Does having blue player in your team makes sense when enemy team has 4 and not a single <50% while your team have four 40% with less than 5k battles.

And show the evidence of what, even if post screenshots of my 10 battles with a winrate gap of 10+ %, you will say its just 10 out of 1000 games. If only blitzstar tracked these sort of  stats then we will know how broken the matchmaking is.

 

View PostRinskiev, on 07 June 2019 - 06:03 AM, said:

I agree with Gandalf. People remember the cases with odd things happening than the usual case. Maybe you didn't understand the answer by WG and they should understand that not everybody knows about statistics: the select by tank type -> players' WR are random -> you may be lucky at times while at others unlucky. Did you check the WR of both teams when you win? OR...is this why I always win?

 

Its not odd occurrence, since 6.0 its almost every next game. Even if I am the one winning its very boring as one-sided games takes the fun out of a multiplayer game. I want fair and competitive matches.

 

 



opavder_13 #11 Posted 08 June 2019 - 12:44 PM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015

View Postpegasoos, on 07 June 2019 - 11:59 AM, said:

 

Ash has always been WG loyal to the bone! I do not know if either he owns a bunch of shares from WG or if he is some kind of [edited] fellow by principle. If he was a good player or had decent stats at least, it would make sense to defending his performance and being able to outplay MM. Or maybe because of being an underperforming player anyway, rigged MM does not make any difference for him?

​MM is very different on various accounts - you surely have noticed that part. The odds of having such loosing strikes (3-4 losses are already statistically odd when you start playing), clearly indicates rigging to incite players to play more, at least 10 games to recover their WR, or for those who do not care about WR, to have a few wins in a session. If you haven't played the game on a certain account for some time, you often get 1-2 Wins for starters, to motivate you to go on! MM is rigged - and that's that!

 

 
 

 

Yeah I have noticed that too. Specially when weekends when there are too many noob players to make these sort of rigging easy.

Newb179 #12 Posted 08 June 2019 - 11:14 PM

    Corporal

  • Member
  • 4827 battles
  • 56
  • [-TN-]
  • Member since:
    07-18-2018
I also agree with Gandalf that Average Win Rate between team Green and team Red will vary only around 3%. I had a 9 consecutive games statistics from December 2018 to prove that, Gandalf had some statistics back from February 2019 to prove that. Do you have a written proof of your allegations? Show some of your consecutive games statistics of yours and enemy teams composition before saying that the match making is skewed.
Now having said that, I shall claim that the match making algorithm is not making a good job of distributing the players. What I mean is there were times when Battle Experience between teams are lopsided, such as team Red had 4 players who had played more than 5000 games while team Green had only 2 players of similar quality with 2 of the least battle experienced players also in team Green. Or vice versa.
I am not being combative or argumentative, but before blaming the match making for your losses please have some written data beforehand. Otherwise it is just hot air.

TnkDztroyer #13 Posted 09 June 2019 - 12:17 AM

    Sergeant

  • Member
  • 7 battles
  • 237
  • Member since:
    02-04-2019

View Postopavder_13, on 08 June 2019 - 10:40 PM, said:

Does having blue player in your team makes sense when enemy team has 4 and not a single <50% while your team have four 40% with less than 5k battles.

And show the evidence of what, even if post screenshots of my 10 battles with a winrate gap of 10+ %, you will say its just 10 out of 1000 games. If only blitzstar tracked these sort of  stats then we will know how broken the matchmaking is.

 

 

Its not odd occurrence, since 6.0 its almost every next game. Even if I am the one winning its very boring as one-sided games takes the fun out of a multiplayer game. I want fair and competitive matches.

 

 

 

ITA. The only good thing about winning steamrolls is that it pushes your winrate up. It's not a satisfying game, or win. I don't mind losing when it's a competitive close game, because then you know you deserved to lose, and the best team one. But those 0/7 steamroll losses are infuriating and do nothing except raise your blood pressure and cause you to punch holes in the wall! (ok, I don't really punch holes in the wall)


When the KV-2 get nerfted, but the Smasher doesn't, we know something is rotten in Denmark. Can anyone say "Pay to Win"?

BossArdnutz #14 Posted 09 June 2019 - 12:05 PM

    Staff Sergeant

  • Member
  • 15392 battles
  • 296
  • [DERPU]
  • Member since:
    03-30-2018
Can't use the skill rating for matchmaking if it's affected by the match results. WR is affected by match results, so you can't use it for matchmaking.

If you try to use the same variable as both input and result of matchmaking, everyone converges on the same, 50/50 value.

erwin10001 #15 Posted 09 June 2019 - 12:20 PM

    First Lieutenant

  • Member
  • 41324 battles
  • 963
  • [NETS]
  • Member since:
    11-12-2016

one big reason for 7-0 losses is that the losing team gets stuck with 2  (or more) inactive/afk players that don't do anything and get 0 dmg. In other words, one team starts off with a 2-tank disadvantage, i.e. 5  v 7.

 

Is it a big surprise than a 2 tank disadvantage translates to a 7-0 loss, excluding the rare occasions where one tank in the disadvantage team manages to pull off a kolobanov and mastery.

 

Its easy for WG to take action against these 0 dmg players, but evidently these players are essential to WG's business model.

 

So you get events that can be completed by 'wins', meaning the player can afk all he wants until he gets the requisite number of wins. If events/missions required a minimum amount of damage in addition to other conditions, I would expect the number of 0 dmg inactive players to be reduced. 

 



opavder_13 #16 Posted 10 June 2019 - 08:40 PM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015
So whats the 

View PostNewb179, on 08 June 2019 - 11:14 PM, said:

I also agree with Gandalf that Average Win Rate between team Green and team Red will vary only around 3%. I had a 9 consecutive games statistics from December 2018 to prove that, Gandalf had some statistics back from February 2019 to prove that. Do you have a written proof of your allegations? Show some of your consecutive games statistics of yours and enemy teams composition before saying that the match making is skewed.
Now having said that, I shall claim that the match making algorithm is not making a good job of distributing the players. What I mean is there were times when Battle Experience between teams are lopsided, such as team Red had 4 players who had played more than 5000 games while team Green had only 2 players of similar quality with 2 of the least battle experienced players also in team Green. Or vice versa.
I am not being combative or argumentative, but before blaming the match making for your losses please have some written data beforehand. Otherwise it is just hot air.

 

I don't see any data to support your claim there's < 3% win rate difference, can you link the data/post here ? also I specifically mentioned version 6.0 in the post so data from december or feb don't do  validate anything.

Its fairly common to see a >7% avg win rate difference like the one below every  3-4 games in 6.0 for me at-least.

 

Whats really stupid is they could have easily balanced it out by simply switching our e100 with there wz-113.

 

 

last match

 

 



Newb179 #17 Posted 11 June 2019 - 12:18 AM

    Corporal

  • Member
  • 4827 battles
  • 56
  • [-TN-]
  • Member since:
    07-18-2018
Okay, that was 1 screenshot. Now do that for 10 games in a row, doesn't matter where you win or lose. Don't forget to add the numbers of battles for each player.
From that screenshot, team green average winning rate is 48%, team red average winning rate is 56%. Population winning rate (team red and green combined) is 52%. So each team average is within 4% of population average.
If you added the battle numbers each player had done, and ranked them from 1 to 14, that would show you how skewed team distribution is from population experience.
I can post my data from December if you can open Excel, but if you want me to provide for 6.0 I have no interest in doing that for version 6.0. Takes too much time.

Newb179 #18 Posted 11 June 2019 - 12:32 AM

    Corporal

  • Member
  • 4827 battles
  • 56
  • [-TN-]
  • Member since:
    07-18-2018
There's simply no mathematical way to predict winning games. There are many times where the more experienced and better winning rate team lost a match because they made mistakes. There's only one truth out there, team who won dealt more damage than the other one.
If you truly want random matchmaking suggest to wargaming like this. When you press that battle button, for example with a tier 5 tank, you are placed on two queues, one with tier 4 and 5, the one with tier 5 and 6.  First one in queue go for green, next one is red. Whichever queue got 14 players first play first. First come first served.

TnkDztroyer #19 Posted 11 June 2019 - 12:32 AM

    Sergeant

  • Member
  • 7 battles
  • 237
  • Member since:
    02-04-2019

View Posterwin10001, on 09 June 2019 - 10:20 PM, said:

one big reason for 7-0 losses is that the losing team gets stuck with 2  (or more) inactive/afk players that don't do anything and get 0 dmg. In other words, one team starts off with a 2-tank disadvantage, i.e. 5  v 7.

 

Is it a big surprise than a 2 tank disadvantage translates to a 7-0 loss, excluding the rare occasions where one tank in the disadvantage team manages to pull off a kolobanov and mastery.

 

Its easy for WG to take action against these 0 dmg players, but evidently these players are essential to WG's business model.

 

So you get events that can be completed by 'wins', meaning the player can afk all he wants until he gets the requisite number of wins. If events/missions required a minimum amount of damage in addition to other conditions, I would expect the number of 0 dmg inactive players to be reduced. 

 

 

The other main reason of course is that one team is so much better than the other.

 

I often have games with one afk, but not usually more. But there are often games where some tanks do very little, which is usually because they get killed quickly due to inexperience, or else they do little because they're afraid of being killed, and that's once again due to inexperience.

 


When the KV-2 get nerfted, but the Smasher doesn't, we know something is rotten in Denmark. Can anyone say "Pay to Win"?

opavder_13 #20 Posted 11 June 2019 - 04:46 AM

    Private

  • Member
  • 24401 battles
  • 8
  • Member since:
    10-31-2015

View PostNewb179, on 11 June 2019 - 12:18 AM, said:

Okay, that was 1 screenshot. Now do that for 10 games in a row, doesn't matter where you win or lose. Don't forget to add the numbers of battles for each player.
From that screenshot, team green average winning rate is 48%, team red average winning rate is 56%. Population winning rate (team red and green combined) is 52%. So each team average is within 4% of population average.
If you added the battle numbers each player had done, and ranked them from 1 to 14, that would show you how skewed team distribution is from population experience.
I can post my data from December if you can open Excel, but if you want me to provide for 6.0 I have no interest in doing that for version 6.0. Takes too much time.

 

I don't need to do a spreadsheet analysis to observe that 3 out of every 5 games are severely skewed in one teams favour. Playing the match and going through the post-battle screen is enough to notice that this matchmaking was retarded and it could have easily made more fair by simply exchanging 1 or 2 players in similar tanks like the example above. The screenshot above was to demonstrate that only. I go through the profiles of both team of every stomped up games, whether i win or loose and this is always the case. So its not just 1 game, so stop pretending to be a smartass by explaining how to do simple statistical analysis.

 

And how do you know that a single sample of size 10 is big & random enough ? that too of consecutive games.

Most regular players plays 500 games a month (15-20 games a day), so 10 will be 2% sample size and mean nothing. 

 

Then theres also the point that why is even a single game has matchmaking like this when it can be corrected by a very simple shuffle of 1-2 players. even if they use a brute force approach of finding best team fit after matchmaking by tank type, then there's simply 2^7 possible combinations. Meaning only 128 shuffles  are needed to get the best matchmaking by win rate out of all possibilities. It will hardly be 25K cpu cycles assuming that 1 shuffle takes 200 cpu cycles which is a very generous estimate (for a well written code it shouldn't take more that 25 cycles). Thats just ~1 ms on a single cpu core, for a calculation done once for every game of 3+ minutes.


Edited by opavder_13, 11 June 2019 - 04:48 AM.






Also tagged with matchmaking

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users