Jump to content


RNG On Accuracy


  • Please log in to reply
13 replies to this topic

TungstenHitman #1 Posted 14 June 2019 - 02:25 PM

    Major General

  • Player
  • 30631 battles
  • 5,765
  • Member since:
    08-28-2016

Hi guys,

 

I have a few questions maybe someone can answer. So, there's an RNG(random numbers generator) in effect in this game with a figure of 25%. So far as I know this applies to accuracy, penetration and damage rolls but I just want to ask about this effect on accuracy in this instance.

 

So, how does this work on accuracy exactly? If we were to run a fair sample size with two tanks, the games most best handling gun, static and sniping fully aimed and the take the games worst handling gun auto aimed on the move at full speed and at maximum turning etc and take 100 shots taken each, my question is.

 

1. Would the moving tank with worst handling gun in the game STILL land 25 shots out of 100 regardless of the laughable bloom, aim time, derpy shell velocity and soft stats?

 

2. Would the static sniper tank with best handling, most accurate gun in the game still miss 25 shots out of 100 regardless of being fully aimed and manually picking out the middle of that target?

 

Or does RNG influence accuracy in others ways than the 25% figure would suggest?



WhoCares01 #2 Posted 14 June 2019 - 02:40 PM

    Warrant Officer

  • Player
  • 21395 battles
  • 741
  • Member since:
    04-21-2015

Others will surely have better information on the subject, but you can look at some analysis that was done between v9.5 and v9.6 (and also 8.5 to 8.6 in a later post) where the handling was supposed to change:

http://forum.worldoftanks.com/index.php?/topic/515227-is-shot-accuracy-rng-the-same-distribution-before-it-is-fully-aimed/

Mind you, that's long before I started playing the game, but I seem to remember discussions about accuracy/dispersion/sigma changes that were immplemented and later reverted/adjusted or some such - as said, others will have more accurate information on the subject.

Overall it seems to follow a bell-shaped distribution function, for which the sigma value defines how steep or flat the curve is, and thus how likely an event will fall close to the center or the edge of the distribution.


Edited by WhoCares01, 14 June 2019 - 02:49 PM.


Kozzy #3 Posted 14 June 2019 - 02:49 PM

    Lieutenant Сolonel

  • Player
  • 42114 battles
  • 3,182
  • [RINSE] RINSE
  • Member since:
    06-29-2011

View PostTungstenHitman, on 14 June 2019 - 01:25 PM, said:

...

 

Or does RNG influence accuracy in others ways than the 25% figure would suggest?

 

As far as I understood it, the 25% is not relevant to accuracy.  For accuracy; the guns have a dispersion diameter and shells will land within this diameter based on a Gaussian distribution.  If plotted on a graph the cross section of the dispersion diameter will look like a bell-curve i.e. more shots tend towards the centre but can go anywhere within the diameter of the reticle.

 

The accuracy of the gun is purely dictated by the dispersion radius (provided that trusty ole WG don't add hidden accuracy stats behind the scenes).

 

This is an old thread which discusses things a bit:

 

http://forum.worldoftanks.eu/index.php?/topic/184204-the-myth-of-accuracy/

 

Another article from the 9.6 change which also explains what is going on:

 

http://ftr.wot-news.com/2015/02/06/9-6-accuracy-change-explained/


Edited by Kozzy, 14 June 2019 - 03:27 PM.


Homer_J #4 Posted 14 June 2019 - 03:13 PM

    Field Marshal

  • Beta Tester
  • 32362 battles
  • 35,391
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostTungstenHitman, on 14 June 2019 - 02:25 PM, said:

Hi guys,

 

I have a few questions maybe someone can answer. So, there's an RNG(random numbers generator) in effect in this game with a figure of 25%. So far as I know this applies to accuracy,

 

I'll stop you there.

 

25% only applies to penetration and damage.

 

Where do people get their info from?



TungstenHitman #5 Posted 14 June 2019 - 03:33 PM

    Major General

  • Player
  • 30631 battles
  • 5,765
  • Member since:
    08-28-2016

View PostHomer_J, on 14 June 2019 - 02:13 PM, said:

 

I'll stop you there.

 

25% only applies to penetration and damage.

 

Where do people get their info from?

 

I don't have the answer to that or why I assumed it applied to accuracy. Is there some sort of WG vid that covers RNG and it applications? I searched on a few occasions but didn't find one on penetration(no joke or referencing the discovery of adult movies) but not any on accuracy and how RNG does or does not apply?

WhoCares01 #6 Posted 14 June 2019 - 03:45 PM

    Warrant Officer

  • Player
  • 21395 battles
  • 741
  • Member since:
    04-21-2015

View PostTungstenHitman, on 14 June 2019 - 03:33 PM, said:

 

I don't have the answer to that or why I assumed it applied to accuracy. Is there some sort of WG vid that covers RNG and it applications? I searched on a few occasions but didn't find one on penetration(no joke or referencing the discovery of adult movies) but not any on accuracy and how RNG does or does not apply?

 

Oh, RNG does apply to accuracy, just not based on the 25%. RNG still decides where your current shot will fall adjusted to the statistical distribution, the distribution just defines how likely it is to hit close or far. And then RNG also decides in which direction your shot deviates, like up, or right, or lower left,or ...

Homer_J #7 Posted 14 June 2019 - 05:08 PM

    Field Marshal

  • Beta Tester
  • 32362 battles
  • 35,391
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostTungstenHitman, on 14 June 2019 - 03:33 PM, said:

 

 not any on accuracy and how RNG does or does not apply?

 

Strange because as Kozzy pointed out there is information out there.

 

There is RNG but 25% doesn't come into it.  Your shot lands within your aim circle with a bias toward the centre.

 

Penetration and damage also have a centre bias but are limited by +/-25%



LCpl_Jones #8 Posted 14 June 2019 - 06:26 PM

    Second Lieutenant

  • Player
  • 13020 battles
  • 1,269
  • [T-D-U] T-D-U
  • Member since:
    06-17-2017

View PostHomer_J, on 14 June 2019 - 05:08 PM, said:

 

........ Your shot lands within your aim circle with a bias toward the centre.......

 

only if Rasha tonk :trollface:

NekoPuffer_PPP #9 Posted 14 June 2019 - 07:37 PM

    Colonel

  • Player
  • 34123 battles
  • 3,985
  • [VRTC] VRTC
  • Member since:
    09-13-2013

View PostLCpl_Jones, on 14 June 2019 - 06:26 PM, said:

View PostHomer_J, on 14 June 2019 - 05:08 PM, said:

........ Your shot lands within your aim circle with a bias toward the centre.......

 

only if Rasha tonk :trollface:

 

Hence the "bias" bit.

Laatikkomafia #10 Posted 14 June 2019 - 07:58 PM

    Major General

  • Beta Tester
  • 23265 battles
  • 5,091
  • [WJDE] WJDE
  • Member since:
    12-27-2010

Accuracy in WoT:

 

 

It is skewed towards the middle. But with better accuracy, the overall is smaller -> the area where shots tend to land becomes smaller.

 

So with a large enough sample size, both the accurate sniper and the derp gun firing on the move will get an expected to-hit chance (depending on the size of the target). Obviously the more accurate sniper gets a larger to-hit percentage.



Dorander #11 Posted 14 June 2019 - 09:19 PM

    Lieutenant General

  • Player
  • 21025 battles
  • 6,031
  • Member since:
    05-07-2012

View PostTungstenHitman, on 14 June 2019 - 01:25 PM, said:

Hi guys,

 

I have a few questions maybe someone can answer. So, there's an RNG(random numbers generator) in effect in this game with a figure of 25%. So far as I know this applies to accuracy, penetration and damage rolls but I just want to ask about this effect on accuracy in this instance.

 

So, how does this work on accuracy exactly? If we were to run a fair sample size with two tanks, the games most best handling gun, static and sniping fully aimed and the take the games worst handling gun auto aimed on the move at full speed and at maximum turning etc and take 100 shots taken each, my question is.

 

1. Would the moving tank with worst handling gun in the game STILL land 25 shots out of 100 regardless of the laughable bloom, aim time, derpy shell velocity and soft stats?

 

2. Would the static sniper tank with best handling, most accurate gun in the game still miss 25 shots out of 100 regardless of being fully aimed and manually picking out the middle of that target?

 

Or does RNG influence accuracy in others ways than the 25% figure would suggest?

 

To add a small bit to Kozzy's explanation, in relation to your questions: If you substitute the 25% accuracy you thought applied with the actual mechanics info Kozzy provided the answer is "yes". You have two aiming circles, and if memory serves two-thirds of all your shots fall inside the inner circle, the remaining one-third somewhere in the outer circle. This applies to every gun, no matter how bad it is. What is different in gun handling is how accuracy (dispersion), bloom and aim-time interact.

 

The short of it is that a tank with poor gun handling, still has 2/3rds of its shot in that inner circle and 1/3rd in outer, but those circles will be bigger. How much bigger depends on the exact stats, if it has high dispersion values the aim circle will be bigger overall, if it has long aim-time the circle shrinks more slowly when you aim so the area where those 2/3rds land will get smaller (as will the 1/3rd outer area), if you have higher bloom values the circle expands more on moving the tank and/or turret. The practical result is that the larger your aim circles are, the more places the shell can land, because you have the same distribution over a larger area.

 

Incidentally shell velocity is not relevant for accuracy, it's only a factor to hit moving targets and how much lead you need to give your shot.



Nishi_Kinuyo #12 Posted 14 June 2019 - 09:25 PM

    Lieutenant General

  • Player
  • 9067 battles
  • 6,259
  • [GUP] GUP
  • Member since:
    05-28-2011

View PostTungstenHitman, on 14 June 2019 - 02:25 PM, said:

2. Would the static sniper tank with best handling, most accurate gun in the game still miss 25 shots out of 100 regardless of being fully aimed and manually picking out the middle of that target?

Numbers might be off, but 4tankersanddog managed to land like 1/3rd of his shots against a stationary target at 200m distance going flat out with a KV-2 derp.


Edited by Nishi_Kinuyo, 14 June 2019 - 09:26 PM.


Homer_J #13 Posted 14 June 2019 - 09:50 PM

    Field Marshal

  • Beta Tester
  • 32362 battles
  • 35,391
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostDorander, on 14 June 2019 - 09:19 PM, said:

 

To add a small bit to Kozzy's explanation, in relation to your questions: If you substitute the 25% accuracy you thought applied with the actual mechanics info Kozzy provided the answer is "yes". You have two aiming circles, and if memory serves two-thirds of all your shots fall inside the inner circle, the remaining one-third somewhere in the outer circle.

 

That's a very simplistic explanation.  Even within your two circles more shots will fall toward the centre.

Dorander #14 Posted 14 June 2019 - 10:19 PM

    Lieutenant General

  • Player
  • 21025 battles
  • 6,031
  • Member since:
    05-07-2012

View PostHomer_J, on 14 June 2019 - 08:50 PM, said:

 

That's a very simplistic explanation.  Even within your two circles more shots will fall toward the centre.

 

Yeah, it was, I was making the point that the same rules apply to all the guns, not discussing the exact functionality of the accuracy formula which as far as I know has never been released.




1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users