Jump to content


Thoughts on Matchmaking and Win Rate

Win Rate Matchmaking random universe

  • Please log in to reply
84 replies to this topic

Eddie_Toms #81 Posted 04 February 2016 - 11:53 PM

    Sergeant

  • Players
  • 53271 battles
  • 310
  • [T0ES]
  • Member since:
    03-01-2015

View PostLephturn, on 04 February 2016 - 04:11 PM, said:

 

The other change they made fixed that - now if there is over 10k people online the tier difference is 0%. Since 2.5 I have not seen a SINGLE game where the total tier differences were not perfectly balanced. They can still improve MM, but this change has been a good one.

 

...

 

I agree the MM change has been very positive in terms of reducing my frustration level.

 

i would like to make a few points.

  • I never said MM (prior to 2.5) was not pseudo random. 
  • I did say the standard deviation in the RNG algorithms appeared to have a large standard deviation
  • These long streaks can be frustrating
  • Besides MM pairings, RNG standard deviation can affect other variables such as spawn position and map selection
  • RNG for ammo hits could also be affected by a large SD 

It's true these are anecdotal observations. Look at my SIG. Look at SIGs of others.

 

Finally, the new MM algorithm shows tank pairing is clearly not the only variable in streaky, frustrating battle experiences. Personally I am less frustrated because it seems to me one source of high SD RNG is minimized or eliminated.


 

 


Mere_Anarchy #82 Posted 05 February 2016 - 04:50 PM

    First Sergeant

  • Players
  • 35681 battles
  • 1,118
  • Member since:
    07-07-2014

View Postneoexodus, on 04 February 2016 - 03:41 PM, said:

VP, thanks for this topic and posts, and I like the idea of the skill plateau that you saw in the initial data set, as that should happen in reality, and lends reliability to your data.  As a new player, I don't have access to this kind of data for my gameplay, but I hope to have it one day, and would bet that it will look pretty similar over the long term.

 

 

You are very welcome. I can't contribute much in terms of game play but like to try and add little bits where I can.  I've managed to nudge up my WR ceiling a bit lately, but it is pretty consistent. I know why too - lack of situational awareness and too much of a tendency to be impatient and take too many hits.

 

View PostLephturn, on 04 February 2016 - 04:11 PM, said:

 

 They can still improve MM, but this change has been a good one.

 

 

Agreed 100%  I am a big fan of the new +1/-1.

 

View PostEddie_Toms, on 04 February 2016 - 05:53 PM, said:

I agree the MM change has been very positive in terms of reducing my frustration level.

 

i would like to make a few points.

  • I never said MM (prior to 2.5) was not pseudo random. 
  • I did say the standard deviation in the RNG algorithms appeared to have a large standard deviation
  • These long streaks can be frustrating
  • Besides MM pairings, RNG standard deviation can affect other variables such as spawn position and map selection
  • RNG for ammo hits could also be affected by a large SD 

It's true these are anecdotal observations. Look at my SIG. Look at SIGs of others.

 

 

The long streaks can indeed be quite demoralizing.  No doubt about it.  I doubt that the new mm will have a large impact on WR overall. For me personally while I enjoy the occasional challenge of trying to carry from 2 tiers down, more often than not it simply doesn't happen. Trying to take on an IS-3/KV-4 with my poor little Fury or Victory just doesn't cut it.  I've enjoyed not having to deal with that frustration.


 

 


The_Motarp #83 Posted 05 February 2016 - 06:41 PM

    Sergeant

  • Players
  • 9143 battles
  • 325
  • Member since:
    11-24-2014
A couple of things, one, MM should never, ever, ever be sorted by player skill, that would penalize good players and reward bad players, which is exactly the opposite of what we want. And second, the people saying that the streaks are too long to be random have no idea what random means. Truly random number generators always produce longer streaks far more often than people trying to be random.

allsgood #84 Posted 14 April 2016 - 05:47 PM

    Lance-corporal

  • Players
  • 99458 battles
  • 47
  • [III-C]
  • Member since:
    06-30-2014

Way to go brother Vanpeenen!   Nicely written piece! Kinda like the stock market. Watching the day to day up and downs will drive you insane. Looking at the long term growth, (or not) is a better guide to your "status!"

 

Thank you!



Mere_Anarchy #85 Posted 15 April 2016 - 05:01 PM

    First Sergeant

  • Players
  • 35681 battles
  • 1,118
  • Member since:
    07-07-2014

View Postallsgood, on 14 April 2016 - 11:47 AM, said:

Way to go brother Vanpeenen!   Nicely written piece! Kinda like the stock market. Watching the day to day up and downs will drive you insane. Looking at the long term growth, (or not) is a better guide to your "status!"

 

Thank you!

 

Thanks man - your kind words are much appreciated!! 

Edited by vanpeenen, 15 April 2016 - 05:54 PM.

 

 





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users