Jump to content


The RNG distribution calculated, dmg + accuracy ( updated)


  • Please log in to reply
121 replies to this topic

brumbarr #1 Posted 11 October 2017 - 12:15 AM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,293
  • [WJDE] WJDE
  • Member since:
    07-30-2012

 

WIth the recent Q&A we got some more info on the dmg and penetration RNG distribution.

So I did some calculations with the info we have from WG:

-normal distribution

-67% to be between -10% and 10%

 

With that I got this distribution:

 

Posted Image

 

 

The x axis is the percentage deviation from the avg roll.

 

With that we can calculate the chance of every roll accuring.

Or if you want to aapt if for 1 particular tank:

Posted Image

 

with sigma= 10,515% of avg dmg

m=avg roll of tank

A=1.017736918

 

 

 

A couple of facts about it:

-37% of shots are between -5% and 5%

-67% of shots are between -10% and 10%

-86% of shots are between -15 and 15%

-96% of shots are between -20% and 20%.

 

Then the accuracy, with this information:

http://ftr.wot-news....ange-explained/

I was able to calculate that pre 9.6 accuracy was a normal distribution with sigma= 0.5225. But in 9.6 they changed it to a customized distribtion that looks like this:

 

Posted Image

 

-28% of shots fall into the  inner 20% radius of the circle.

-51% of shots fall into the  inner 35% radius of the circle.

-70%  of shots fall into the  inner 50% radius of the circle.

-87% of shots fall into the inner 70% radius of the circle.

-97% of shots fall into the inner 90% radius of the circle.

 

 

 


Edited by brumbarr, 14 October 2017 - 08:16 PM.


IncandescentGerbil #2 Posted 11 October 2017 - 12:34 AM

    Second Lieutenant

  • Player
  • 35597 battles
  • 1,443
  • Member since:
    11-24-2015
I can't take any of this seriously when your opening paragraph contains a comma splice.

K_A #3 Posted 11 October 2017 - 12:39 AM

    Brigadier

  • Player
  • 13566 battles
  • 4,656
  • [T-D-U] T-D-U
  • Member since:
    04-18-2013
I distinctly remember a fair number of times I've rolled 293 on Russian 122mm guns, but then of course I'll remember them because they are highly unlikely and lowest possible figure.

brumbarr #4 Posted 11 October 2017 - 12:45 AM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,293
  • [WJDE] WJDE
  • Member since:
    07-30-2012

View PostIncandescentGerbil, on 11 October 2017 - 12:34 AM, said:

I can't take any of this seriously when your opening paragraph contains a comma splice.

 

Maths isnt grammar, 

and tbh, I dont even know what a comma splice is ;)



IncandescentGerbil #5 Posted 11 October 2017 - 12:52 AM

    Second Lieutenant

  • Player
  • 35597 battles
  • 1,443
  • Member since:
    11-24-2015

View Postbrumbarr, on 11 October 2017 - 12:45 AM, said:

 

Maths isnt grammar, 

and tbh, I dont even know what a comma splice is ;)

Don't sweat it dude. Your equations look incredibly sexy, which is all that matters.



Velvet_Underground #6 Posted 11 October 2017 - 12:55 AM

    Lieutenant Сolonel

  • Player
  • 22243 battles
  • 3,164
  • Member since:
    12-19-2014

View Postbrumbarr, on 11 October 2017 - 12:15 AM, said:

I know very few people will care or are intelligent anough enough to get what I will write in this forum, there is prob a better place for it. But anyway:

:hiding:



brumbarr #7 Posted 11 October 2017 - 01:12 AM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,293
  • [WJDE] WJDE
  • Member since:
    07-30-2012

View PostVelvet_Underground, on 11 October 2017 - 12:55 AM, said:

:hiding:

 

oh ffs :P

_EXODUZ_ #8 Posted 11 October 2017 - 01:37 AM

    Lieutenant

  • Player
  • 34165 battles
  • 1,935
  • [RGT] RGT
  • Member since:
    11-05-2014

View Postbrumbarr, on 11 October 2017 - 12:15 AM, said:

...what I will write

 

View Postbrumbarr, on 11 October 2017 - 01:12 AM, said:

 

oh ffs :P

 

Had to do it, sorry. :izmena:

KillingJoker #9 Posted 11 October 2017 - 02:20 AM

    Second Lieutenant

  • Player
  • 28958 battles
  • 1,149
  • Member since:
    09-07-2015

View Postbrumbarr, on 11 October 2017 - 12:15 AM, said:

A couple of facts about it:

-37% of shots are between -5% and 5%

-67% of shots are between -10% and 10%

-86% of shots are between -15 and 15%

-96% of shots are between -20% and 20%.

 

 

So you mean, that if play my IS-4, with the 122 mm gun that have 258 mm of penetration

only 37% of the time i shot the gun the shell penetration will have more or less the values expected (258) ?

 

would be very interesting, if there was a mod to track the shell penetration on targets, just to know exacly what was the force of the impact of the shell that either pened or bounced

 

 


Edited by KillingJoker, 11 October 2017 - 02:21 AM.


ZlatanArKung #10 Posted 11 October 2017 - 06:49 AM

    Major General

  • Player
  • 1529 battles
  • 5,112
  • Member since:
    12-20-2014

View PostKillingJoker, on 11 October 2017 - 02:20 AM, said:

 

 

So you mean, that if play my IS-4, with the 122 mm gun that have 258 mm of penetration

only 37% of the time i shot the gun the shell penetration will have more or less the values expected (258) ?

 

would be very interesting, if there was a mod to track the shell penetration on targets, just to know exacly what was the force of the impact of the shell that either pened or bounced

 

 

 

Same distribution for damage afaik.

And damage is easier to track.



NUKLEAR_SLUG #11 Posted 11 October 2017 - 07:45 AM

    Lieutenant

  • Player
  • 26628 battles
  • 1,619
  • Member since:
    06-13-2015
I think this is about the point someone rolls in to claim their games are rigged to always give them values at the wrong end of the curve..

Geno1isme #12 Posted 11 October 2017 - 09:17 AM

    Lieutenant General

  • Player
  • 40053 battles
  • 6,629
  • [TRYIT] TRYIT
  • Member since:
    09-03-2013

View PostZlatanArKung, on 11 October 2017 - 07:49 AM, said:

And damage is easier to track.

 

Or rather damage is the only RNG-based value you have a chance to track to some degree, as for hitting and penning the server will only tell you a yes/no response, not the actual values involved. And even if you'd get the actual pen value involved there is still distance falloff and normalization to account for, not to mention the effective armor value.

Edited by Geno1isme, 11 October 2017 - 09:19 AM.


Spurtung #13 Posted 11 October 2017 - 09:27 AM

    Major General

  • Player
  • 60137 battles
  • 5,519
  • [GW-UP] GW-UP
  • Member since:
    07-05-2013

View Postbrumbarr, on 11 October 2017 - 01:15 AM, said:

A couple of facts about it:

-37% of shots are between -5% and 5%

-67% of shots are between -10% and 10%

-86% of shots are between -15 and 15%

-96% of shots are between -20% and 20%.

2/3 of shots with a +/-10% is pretty good.



G01ngToxicCommand0 #14 Posted 11 October 2017 - 09:35 AM

    Warrant Officer

  • Player
  • 37465 battles
  • 727
  • Member since:
    11-03-2011

View Postbrumbarr, on 11 October 2017 - 12:15 AM, said:

I know very few people will care or are intelligent anough to get what I wil write in this forum, there is prob a better place for it. But anyway:

WIth the recent Q&A we got some more info on the dmg and penetration RNG distribution.

So I did some calculations with the info we have from WG:

-normal distribution

-67% to be between -10% and 10%

 

With that I got this distribution:

 

Posted Image

 

The x axis is the percentage deviation from the avg roll.

 

With that we can calculate the chance of every roll accuring.

Or if you want to aapt if for 1 particular tank:

Posted Image

 

with sigma= 10,515% of avg dmg

m=avg roll of tank

A=1.017736918

 

 

 

A couple of facts about it:

-37% of shots are between -5% and 5%

-67% of shots are between -10% and 10%

-86% of shots are between -15 and 15%

-96% of shots are between -20% and 20%.

 

Or the RNG is a just a large table containing cell X0 to Xn with discrete values set up to simulate a mathematical formula with a randomiser that has the same probability to pick cell X0 as cell Xn which at the game interface would be completely indistinguishable from an advanced gaussian distribution formula.

Edited by G01ngToxicCommand0, 11 October 2017 - 09:37 AM.


Dr_ownape #15 Posted 11 October 2017 - 09:52 AM

    Major General

  • Player
  • 40642 battles
  • 5,146
  • [IDEAL] IDEAL
  • Member since:
    03-27-2013
why did i come here, now i need a rest :(

Long_Range_Sniper #16 Posted 11 October 2017 - 09:53 AM

    Lieutenant General

  • Player
  • 29462 battles
  • 7,543
  • [-MM] -MM
  • Member since:
    04-04-2011

Which all lets WG change RNG around a normal distribution, and yet still keep the majority around the median line, but increase the number of shots that could miss.

 

Image result for normal distribution gif



eldrak #17 Posted 11 October 2017 - 09:54 AM

    Warrant Officer

  • Player
  • 45844 battles
  • 975
  • [NE-VO] NE-VO
  • Member since:
    01-27-2011

View PostG01ngToxicCommand0, on 11 October 2017 - 08:35 AM, said:

 

Or the RNG is a just a large table containing cell X0 to Xn with discrete values set up to simulate a mathematical formula with a randomiser that has the same probability to pick cell X0 as cell Xn which at the game interface would be completely indistinguishable from an advanced gaussian distribution formula.

 

That would be less practical due to differing damage values for guns.

 

Brumbarr, could you integrate the values from +25 to inf (or -25 to -inf) to show what the true chance of doing max or min rolls are.

(Doing 1 more or less than min/max is quite uncommon while minmax rolls happen relatively often, in my experience)

 

 



mikedee #18 Posted 11 October 2017 - 10:28 AM

    Lance-corporal

  • Beta Tester
  • 20326 battles
  • 72
  • Member since:
    07-16-2010

View Postbrumbarr, on 10 October 2017 - 11:15 PM, said:

I know very few people will care or are intelligent anough to get what I wil write in this forum, 

 

:rolleyes:

Yes, in the year of our lord of 2017, only the first borns of barons have the privilege to experience 12th grade maths.

Anyway, your effort is appreciated.



brumbarr #19 Posted 11 October 2017 - 10:39 AM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,293
  • [WJDE] WJDE
  • Member since:
    07-30-2012

View Postmikedee, on 11 October 2017 - 10:28 AM, said:

 

:rolleyes:

Yes, in the year of our lord of 2017, only the first borns of barons have the privilege to experience 12th grade maths.

Anyway, your effort is appreciated.

This is the wot forums, reading this everyday you lose hope in humanity.



jabster #20 Posted 11 October 2017 - 10:43 AM

    Field Marshal

  • Beta Tester
  • 12516 battles
  • 21,730
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View Posteldrak, on 11 October 2017 - 08:54 AM, said:

 

That would be less practical due to differing damage values for guns.

 

 

Not really no as all you need is a table with the correct distribution of -25% to +25% values and you then use that percentage to modify the base damage. Not that I think it makes any real difference, to the user, which ever way it's done.




1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users