Jump to content


Leo PTA Low roll bais?


  • Please log in to reply
30 replies to this topic

Schemezoid #1 Posted 26 July 2017 - 10:09 PM

    Major

  • Player
  • 20128 battles
  • 2,710
  • Member since:
    11-29-2013

Hello people!

Prolly not the most original thread around, but i was wondering....

I just unlocked the Leo PTA. Pretty nice tank if you ask me.

 

But the gun, oh the gun!

I noticed that the gun seemed to low roll really often. So instead of going to the forum to make a complaint thread i first gathered evidence.

We all know how you forumites love numbers, statistics and trolling.

 

So here we go!

 

Ive gathered the damage rolls of my last 100 shots fired. Seems like a fair amount, right? RIGHT?

 

-23 fell in the +20/-20 dam bracket, which is nice

-51 shots did less then 370dam

-26 shots did more then 410dam

SO over half the shots did noticable less damage then the "average" roll of 390

 

So my question here for you!

Is this gun behaving the same for everyone? Or did i get targeted by Wg's all so infamous "negative RNG formula"? :bush:

 

The only other gun that really rolls low on average is the T34's gun. But gathering that data would require me to play the T34 again, so no!.

 

Anyways, enlighten me!

 

 



Poerhis #2 Posted 26 July 2017 - 10:33 PM

    Sergeant

  • Beta Tester
  • 16654 battles
  • 209
  • [SIKA] SIKA
  • Member since:
    01-14-2011

Just bad luck. Over time it's very likely that the average damage sets near 390. A hundred shots is not really a sufficient quantitive amount to draw conclusions on.

 

Lovely tank by the way. :)



Schemezoid #3 Posted 26 July 2017 - 10:34 PM

    Major

  • Player
  • 20128 battles
  • 2,710
  • Member since:
    11-29-2013

View PostIllort, on 26 July 2017 - 09:33 PM, said:

Just bad luck. Over time it's very likely that the average damage sets near 390. A hundred shots is not really a sufficient quantitive amount to draw conclusions on.

 

Lovely tank by the way. :)

 

Its a amazing tank. Feels so much more responsive then the AMX30PT.

 

Guess ill keep adding numbers then, we'll see ;)



Poerhis #4 Posted 26 July 2017 - 10:37 PM

    Sergeant

  • Beta Tester
  • 16654 battles
  • 209
  • [SIKA] SIKA
  • Member since:
    01-14-2011

Would be interesting to see a long term Excel sheet on the subject! Gotta remove all killshots and HE hits though. :)

 

You seem to be doing well in the tank. Glad you enjoy it. It's one of my favorites, very near to three marking it over here. :)



Schemezoid #5 Posted 26 July 2017 - 10:40 PM

    Major

  • Player
  • 20128 battles
  • 2,710
  • Member since:
    11-29-2013
Only counting Apcr/heat. I wonder if killshots are rolled aswell. Because imagine if you killed 30 vehicles and 20 of those "would have highrolled", then the sheet looks different alrdy.

DracheimFlug #6 Posted 26 July 2017 - 10:50 PM

    Colonel

  • Player
  • 8925 battles
  • 3,725
  • Member since:
    11-13-2014
Eh, the parent teacher associations of other animals have low roll calls, too. Except fish. They are big on schools... 

Homer_J #7 Posted 26 July 2017 - 11:14 PM

    Field Marshal

  • Beta Tester
  • 27072 battles
  • 27,725
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostSchemezoid, on 26 July 2017 - 10:40 PM, said:

I wonder if killshots are rolled aswell.

 

Of course they are or you could never get a low roll none kill shot which would have killed on an average roll.

 

BTW don't you need to exclude low rolls which should have killed to balance out the high roll kill shots which you are excluding?

 

i.e. of all the shots against a 390 hp target you are only counting the low rolls, so aren't you setting it up to get more low rolls in your stats?

 

Yep, I'm convinced you should only count shots against enemies with at least 488 hitpoints and exclude ammo rack kills.


Edited by Homer_J, 26 July 2017 - 11:20 PM.


BabyR4ge #8 Posted 27 July 2017 - 01:10 AM

    Private

  • Beta Tester
  • 28419 battles
  • 44
  • [GSQT] GSQT
  • Member since:
    10-05-2010
You meant "bias" op ....

jabster #9 Posted 27 July 2017 - 04:36 AM

    Field Marshal

  • Beta Tester
  • 12516 battles
  • 21,707
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostHomer_J, on 26 July 2017 - 10:14 PM, said:

 

Of course they are or you could never get a low roll none kill shot which would have killed on an average roll.

 

BTW don't you need to exclude low rolls which should have killed to balance out the high roll kill shots which you are excluding?

 

i.e. of all the shots against a 390 hp target you are only counting the low rolls, so aren't you setting it up to get more low rolls in your stats?

 

Yep, I'm convinced you should only count shots against enemies with at least 488 hitpoints and exclude ammo rack kills.

 

Basically yes. You have to exclude all shots that don't have the potential to have a max. roll otherwise, as you say, you bias your results towards low values. It sounds a bit strange until you think about having one hundred shots against a target that has 1% more hit points than your average roll. If you just excluded kill shots then your average damage roll is going to be way off.

 

I'm not saying that's what the OP has done though.



Phobos4321 #10 Posted 27 July 2017 - 05:59 AM

    Lieutenant General

  • Player
  • 42827 battles
  • 7,828
  • Member since:
    09-27-2011

also the problem would come from ignoring shots

since no one could tell if that 30 damage kill shot was for example a high/max roll

 

so even every killshot below 390 damage could be actually a highroll missing in the statistics



DracheimFlug #11 Posted 27 July 2017 - 07:21 AM

    Colonel

  • Player
  • 8925 battles
  • 3,725
  • Member since:
    11-13-2014

View Postjabster, on 27 July 2017 - 04:36 AM, said:

 

Basically yes. You have to exclude all shots that don't have the potential to have a max. roll otherwise, as you say, you bias your results towards low values. It sounds a bit strange until you think about having one hundred shots against a target that has 1% more hit points than your average roll. If you just excluded kill shots then your average damage roll is going to be way off.

 

I'm not saying that's what the OP has done though.

 

This.

 

View PostPhobos4321, on 27 July 2017 - 05:59 AM, said:

also the problem would come from ignoring shots

since no one could tell if that 30 damage kill shot was for example a high/max roll

 

so even every killshot below 390 damage could be actually a highroll missing in the statistics

 

Not this. Assuming each shot is properly independent, then excluding the kill shots should be no different than ignoring any given set of rolls. Ignoring a thousand kill shots should give similar results to ignoring the thousand shots before you started tracking shots, or the thousand shots after you stop.



Baldrickk #12 Posted 27 July 2017 - 07:30 AM

    Field Marshal

  • Player
  • 29031 battles
  • 12,882
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View Postjabster, on 27 July 2017 - 04:36 AM, said:

 

Basically yes. You have to exclude all shots that don't have the potential to have a max. roll otherwise, as you say, you bias your results towards low values. It sounds a bit strange until you think about having one hundred shots against a target that has 1% more hit points than your average roll. If you just excluded kill shots then your average damage roll is going to be way off.

 

I'm not saying that's what the OP has done though.

If  you are only after number, not magnitude, you only have to limit it to shots on targets where HP>avDmg.

 

Recording magnitude would be useful though. You could check that there is indeed a Gaussian distribution around the average



Schemezoid #13 Posted 27 July 2017 - 07:37 AM

    Major

  • Player
  • 20128 battles
  • 2,710
  • Member since:
    11-29-2013
Is there a mod that records shots fired and the damage they did?
Would be nice to record a greater amount of numbers. A thousand shots should eliminate the offset you get by using smaller numbers.

Baldrickk #14 Posted 27 July 2017 - 07:40 AM

    Field Marshal

  • Player
  • 29031 battles
  • 12,882
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View PostSchemezoid, on 27 July 2017 - 07:37 AM, said:

Is there a mod that records shots fired and the damage they did?
Would be nice to record a greater amount of numbers. A thousand shots should eliminate the offset you get by using smaller numbers.

As xvm records the numbers, you could probably get it to export them.

But you might have trouble with the whole enemy hp thing



jabster #15 Posted 27 July 2017 - 09:16 AM

    Field Marshal

  • Beta Tester
  • 12516 battles
  • 21,707
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostBaldrickk, on 27 July 2017 - 06:30 AM, said:

If  you are only after number, not magnitude, you only have to limit it to shots on targets where HP>avDmg.

 

Recording magnitude would be useful though. You could check that there is indeed a Gaussian distribution around the average

 

So only if you want to do it properly then?

Enforcer1975 #16 Posted 27 July 2017 - 12:07 PM

    General

  • Player
  • 18425 battles
  • 9,847
  • [WJDE] WJDE
  • Member since:
    05-04-2014
You know what came to mind when i read the OP? You get rng on damage right?
So if you hit a tank it's in case of the PTA 390 damage +- 25%
If the shooter rolls higher it could be 400 but why is the average damage counted blocked and not what the shot could potentially have done? Not that it matters because blocked is blocked but yeah...pedantry...

Homer_J #17 Posted 27 July 2017 - 04:54 PM

    Field Marshal

  • Beta Tester
  • 27072 battles
  • 27,725
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostEnforcer1975, on 27 July 2017 - 12:07 PM, said:

You know what came to mind when i read the OP? You get rng on damage right?
So if you hit a tank it's in case of the PTA 390 damage +- 25%
If the shooter rolls higher it could be 400 but why is the average damage counted blocked and not what the shot could potentially have done? Not that it matters because blocked is blocked but yeah...pedantry...

 

It seems that the damage roll for AP/APCR/HEAT is only calculated after the game decides you have penetrated.  Which makes sense from a programming point of view, no point in committing CPU cycles to something you might not need.  A side effect is that damage blocked is always the average.



Ulfarr #18 Posted 27 July 2017 - 05:56 PM

    Sergeant

  • Player
  • 26862 battles
  • 267
  • Member since:
    04-07-2012

View Postjabster, on 27 July 2017 - 05:36 AM, said:

 

Basically yes. You have to exclude all shots that don't have the potential to have a max. roll otherwise, as you say, you bias your results towards low values. It sounds a bit strange until you think about having one hundred shots against a target that has 1% more hit points than your average roll. If you just excluded kill shots then your average damage roll is going to be way off.

 

I'm not saying that's what the OP has done though.

 

View PostHomer_J, on 27 July 2017 - 12:14 AM, said:

 

Of course they are or you could never get a low roll none kill shot which would have killed on an average roll.

 

BTW don't you need to exclude low rolls which should have killed to balance out the high roll kill shots which you are excluding?

 

i.e. of all the shots against a 390 hp target you are only counting the low rolls, so aren't you setting it up to get more low rolls in your stats?

 

Yep, I'm convinced you should only count shots against enemies with at least 488 hitpoints and exclude ammo rack kills.

 

 

I might be misunderstanding what you two are saying, but I don't think that this is neccesary. The only shots that should be excluded are killing shots (and ammoracks), as from their very nature, the player can not know their exact number and if they were a max roll or not. Any other shot, that fails to kill, should be eligible since there would be no lost potential.

 

The problem of biasing the results to show lower numbers could be effectively mitigated by increasing the sample size (ie 1000 shots)  since then the killing shots would be too few* to have any significant impact.

 

 

*important edit: That is unless the player tries to remove tanks from the match specifically thus focusing mainly on hitting targets that can be one shot.


Edited by Ulfarr, 27 July 2017 - 06:03 PM.


jabster #19 Posted 27 July 2017 - 06:10 PM

    Field Marshal

  • Beta Tester
  • 12516 battles
  • 21,707
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostUlfarr, on 27 July 2017 - 04:56 PM, said:

 

 

 

I might be misunderstanding what you two are saying, but I don't think that this is neccesary. The only shots that should be excluded are killing shots (and ammoracks), as from their very nature, the player can not know their exact number and if they were a max roll or not. Any other shot, that fails to kill, should be eligible since there would be no lost potential.

 

The problem of biasing the results to show lower numbers could be effectively mitigated by increasing the sample size (ie 1000 shots)  since then the killing shots would be too few* to have any significant impact.

 

 

*important edit: That is unless the player tries to remove tanks from the match specifically thus focusing mainly on hitting targets that can be one shot.

 

You fire on a target with 100 hp with and a gun that does an average of 100 damage. If you only discard killshots then your results will show you never  roll above average. 

Ulfarr #20 Posted 27 July 2017 - 06:38 PM

    Sergeant

  • Player
  • 26862 battles
  • 267
  • Member since:
    04-07-2012

View Postjabster, on 27 July 2017 - 07:10 PM, said:

 

You fire on a target with 100 hp with and a gun that does an average of 100 damage. If you only discard killshots then your results will show you never  roll above average. 

 

 

In that case, yes. But that's because *this* experiment would remove any possibility to roll high. 

 

However that is not what I'm arguing about. You claim that Schemezoid should exclude any shot that was made on targets with less than 488 hp, but that seems unneccesary.  For example, a 480 dmg shot on a 481 hp target can't bias the results into showing lower numbers since there is no lost damage potential.






1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users