Jump to content


Alpha damage spread/RNG

rng russian bias alpha sigma

  • Please log in to reply
33 replies to this topic

GrapefruitSalad #1 Posted 03 July 2020 - 04:27 PM

    Corporal

  • Player
  • 58672 battles
  • 181
  • [CELL] CELL
  • Member since:
    04-14-2012

Hi all,

 

It is no secret anymore that WG's accuracy numbers are complete lies, with so called "sigma" affecting the spread of where shells go in the aim circle.. i.e. KanJpz 105 and Skorp G have very good on paper "accuracies", but the sigma screws them over my making the shells fly all over the small aiming circle. On the other hand, tanks like the M46 have bad accuracy, but very high sigma so that shells tend to go straight to the middle of the relatively large aim circle.

 

This is all very obvious for people who have played plenty of games, particularly in these tanks.

 

However, I was making this thread to highlight the alpha damage "sigma", where a tank's on paper average damage, is not always the case. How many times have you been killed by two high rolls from a Russian tank. As it's lockdown and playing this game is too frustrating anyway, I watched many of my replays in various tanks to see if I just have my tin foil hat on, or if there really is something odd going on. 

 

Unfortunately, for the majority of tanks, there was nothing that weird, however here are some strange results I found:

 

 

Skoda T27:

 

Sample: 40 battles

Average damage: 240

Actual average damage: 223.6

 

 

AMX 50 Foch 155:

 

Sample: 23 battles

Average damage: 750

Actual average damage: 703.0

 

 

Leopard 1:

 

Sample: 28 battles

Average damage: 420

Actual average damage: 435.1

 

 

Object 416:

 

Sample: 30 battles

Average damage: 320

Actual average damage: 333.2

 

 

Maybe my sample size is not big enough but I couldn't be bothered to do any more. If I had overpowered Russian tanks like the Obj 277 and IS-7, I would be interested to see there result as it is very rare to see an IS-7 roll below 500.

 

What are your thoughts on this, do you think there is something fishy here? Any tanks in particular you think tend to low roll/high roll all the time?

 

 

(Btw this is not a Russian bias post but I'm pretty sure you'd find most Russian 490 guns roll for over 500 most of the time, and Russian 440 guns always kill in one shot tanks with 480HP)

 

Ps. If WG delete this post, we can confirm it is true



Erwin_Von_Braun #2 Posted 03 July 2020 - 04:33 PM

    Lieutenant General

  • Player
  • 46920 battles
  • 7,564
  • Member since:
    01-25-2014

Cue the tirade of fully qualified data analysts willing to dispute anything that dares to contradict established dogma...

 



ZlatanArKung #3 Posted 03 July 2020 - 04:37 PM

    Lieutenant General

  • Player
  • 1537 battles
  • 6,177
  • Member since:
    12-20-2014

View PostGrapefruitSalad, on 03 July 2020 - 04:27 PM, said:

Hi all,

 

It is no secret anymore that WG's accuracy numbers are complete lies, with so called "sigma" affecting the spread of where shells go in the aim circle.. i.e. KanJpz 105 and Skorp G have very good on paper "accuracies", but the sigma screws them over my making the shells fly all over the small aiming circle. On the other hand, tanks like the M46 have bad accuracy, but very high sigma so that shells tend to go straight to the middle of the relatively large aim circle.

 

This is all very obvious for people who have played plenty of games, particularly in these tanks.

 

However, I was making this thread to highlight the alpha damage "sigma", where a tank's on paper average damage, is not always the case. How many times have you been killed by two high rolls from a Russian tank. As it's lockdown and playing this game is too frustrating anyway, I watched many of my replays in various tanks to see if I just have my tin foil hat on, or if there really is something odd going on. 

 

Unfortunately, for the majority of tanks, there was nothing that weird, however here are some strange results I found:

 

 

Skoda T27:

 

Sample: 40 battles

Average damage: 240

Actual average damage: 223.6

 

 

AMX 50 Foch 155:

 

Sample: 23 battles

Average damage: 750

Actual average damage: 703.0

 

 

Leopard 1:

 

Sample: 28 battles

Average damage: 420

Actual average damage: 435.1

 

 

Object 416:

 

Sample: 30 battles

Average damage: 320

Actual average damage: 333.2

 

 

Maybe my sample size is not big enough but I couldn't be bothered to do any more. If I had overpowered Russian tanks like the Obj 277 and IS-7, I would be interested to see there result as it is very rare to see an IS-7 roll below 500.

 

What are your thoughts on this, do you think there is something fishy here? Any tanks in particular you think tend to low roll/high roll all the time?

 

 

(Btw this is not a Russian bias post but I'm pretty sure you'd find most Russian 490 guns roll for over 500 most of the time, and Russian 440 guns always kill in one shot tanks with 480HP)

 

Ps. If WG delete this post, we can confirm it is true

Do you know what a sigma value is?



BattleMetalChris #4 Posted 03 July 2020 - 04:41 PM

    General

  • Player
  • 11269 battles
  • 8,016
  • [WJDE] WJDE
  • Member since:
    05-18-2011
Is that dis-counting kill-shots?

If you shoot a tank and kill it, you can only deal damage up to its remaining hitpoints, which is likely to be lower than the gun's average damage. You need to remove kill-shots from your sample.

Edited by BattleMetalChris, 03 July 2020 - 04:41 PM.


GodTank2 #5 Posted 03 July 2020 - 04:41 PM

    Warrant Officer

  • Player
  • 11541 battles
  • 995
  • [RDDT] RDDT
  • Member since:
    12-14-2012

Leopard 1 is russian confirmed?

58K games and you sit there and believe this crapreally?



Gruff_ #6 Posted 03 July 2020 - 04:46 PM

    Major

  • Player
  • 26383 battles
  • 2,749
  • Member since:
    09-22-2011
Very small sample size to show much variance from the predicted average damage.

ZlatanArKung #7 Posted 03 July 2020 - 04:47 PM

    Lieutenant General

  • Player
  • 1537 battles
  • 6,177
  • Member since:
    12-20-2014

View PostErwin_Von_Braun, on 03 July 2020 - 04:33 PM, said:

Cue the tirade of fully qualified data analysts willing to dispute anything that dares to contradict established dogma...

 

Tin foil hat brigade suits you well.



Erwin_Von_Braun #8 Posted 03 July 2020 - 04:53 PM

    Lieutenant General

  • Player
  • 46920 battles
  • 7,564
  • Member since:
    01-25-2014

View PostZlatanArKung, on 03 July 2020 - 03:47 PM, said:

Tin foil hat brigade suits you well.

Where did I say that I disagreed?

Or was I merely pointing out the inherent predictability of certain forum types?

 



jabster #9 Posted 03 July 2020 - 04:57 PM

    Field Marshal

  • Beta Tester
  • 12856 battles
  • 28,290
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostBattleMetalChris, on 03 July 2020 - 03:41 PM, said:

Is that dis-counting kill-shots?

If you shoot a tank and kill it, you can only deal damage up to its remaining hitpoints, which is likely to be lower than the gun's average damage. You need to remove kill-shots from your sample.


It’s not just kill shots, you need to remove all shots that couldn’t have done maximum damage.



GrapefruitSalad #10 Posted 03 July 2020 - 05:20 PM

    Corporal

  • Player
  • 58672 battles
  • 181
  • [CELL] CELL
  • Member since:
    04-14-2012

View PostGodTank2, on 03 July 2020 - 03:41 PM, said:

Leopard 1 is russian confirmed?

58K games and you sit there and believe this crapreally?


58k games of getting high rolled by Defenders



shikaka9 #11 Posted 03 July 2020 - 05:23 PM

    Major

  • Player
  • 80039 battles
  • 2,606
  • Member since:
    02-27-2013
burning damage counted or not?

GrapefruitSalad #12 Posted 03 July 2020 - 05:33 PM

    Corporal

  • Player
  • 58672 battles
  • 181
  • [CELL] CELL
  • Member since:
    04-14-2012

View Postshikaka9, on 03 July 2020 - 04:23 PM, said:

burning damage counted or not?


nope. Only full damage shots, aka no fires and no killing shots



jabster #13 Posted 03 July 2020 - 05:43 PM

    Field Marshal

  • Beta Tester
  • 12856 battles
  • 28,290
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostGrapefruitSalad, on 03 July 2020 - 04:33 PM, said:


nope. Only full damage shots, aka no fires and no killing shots


But did you include shots that didn’t have the potential for a max roll as that will bias the average damage downwards.



NUKLEAR_SLUG #14 Posted 03 July 2020 - 05:48 PM

    Lieutenant General

  • Player
  • 37491 battles
  • 6,525
  • [FISHY] FISHY
  • Member since:
    06-13-2015

View PostGrapefruitSalad, on 03 July 2020 - 04:27 PM, said:

It is no secret anymore that WG's accuracy numbers are complete lies, with so called "sigma" affecting the spread of where shells go in the aim circle..

 

You should stop listening to Klaus. 



Spurtung #15 Posted 03 July 2020 - 05:53 PM

    General

  • Player
  • 82711 battles
  • 8,499
  • [USSRX] USSRX
  • Member since:
    07-05-2013

View PostGrapefruitSalad, on 03 July 2020 - 03:27 PM, said:

so called "sigma"

 

Showing how ignorant you are right on the first line. That's where anyone with over half a brain stopped reading.



GrapefruitSalad #16 Posted 03 July 2020 - 05:56 PM

    Corporal

  • Player
  • 58672 battles
  • 181
  • [CELL] CELL
  • Member since:
    04-14-2012

View Postjabster, on 03 July 2020 - 04:43 PM, said:


But did you include shots that didn’t have the potential for a max roll as that will bias the average damage downwards.


I did include them, although I can see now how that could skew things. Good point 

16:58 Added after 1 minute

View PostSpurtung, on 03 July 2020 - 04:53 PM, said:

 

Showing how ignorant you are right on the first line. That's where anyone with over half a brain stopped reading.


that’s why you continued reading, arty player



Private_Miros #17 Posted 03 July 2020 - 06:02 PM

    Field Marshal

  • Player
  • 30063 battles
  • 11,948
  • [EMU87] EMU87
  • Member since:
    07-09-2011

Those remarkable results, I also kept mine for a while, these are tanks I recorded 50 or more shots for (only counting those on targets with more HP than the maximum roll, and only those I could verify in Battle Hits after the battle - so no blind shots either):

 

Observed Average Tank
245,67 240 Panther 88
135,47 135 E 25
85,29 85 T-50-2
178,66 180 Rudy USSR
180,44 180 T-34-85M
395,13 390 SU-100
117,07 115 T37
238,47 240 T20


Dava_117 #18 Posted 03 July 2020 - 06:03 PM

    Lieutenant General

  • Moderator
  • 24696 battles
  • 6,967
  • [T-D-U] T-D-U
  • Member since:
    12-17-2014

The First part of you OP is false as the sigma is the same for all tanks. WG even confirmed they calculate arrays of dispersion points, during low load times, and then apply them in sequence once someone fire the gun, adapting the calculated point on the reticule of the shooter.



DaniulSims #19 Posted 03 July 2020 - 06:06 PM

    Second Lieutenant

  • Player
  • 11477 battles
  • 1,199
  • [ELC3K] ELC3K
  • Member since:
    03-29-2014
I'm rather positive the OP isn't aware of what Sigma is.

24doom24 #20 Posted 03 July 2020 - 06:16 PM

    Captain

  • Player
  • 9571 battles
  • 2,068
  • [WW3] WW3
  • Member since:
    10-20-2012

Private Miros data: 

Posted Image

Mean: 85.29

Median: 86

Mode: 89 (18)

Standard Deviation: 7.2757


Edited by 24doom24, 03 July 2020 - 06:20 PM.






Also tagged with rng, russian bias, alpha, sigma

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users