Jump to content


Tank OP ratings


  • Please log in to reply
103 replies to this topic

Baldrickk #1 Posted 30 July 2017 - 06:34 PM

    Field Marshal

  • Player
  • 30142 battles
  • 14,301
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

Hi all.

 

I've started working on a small program to try and provide an OP-ness score for tanks, based on their recent WR curves.

 

The number is the difference between the integration of the best fit line for the availiable data between 45% and 65%, less the integration of the reference curve across the same range.

 

The idea was to create a rating that would show a strong positive indicator if a tank performed well for most players, and likewise for underperforming tanks, it would receive a negative rating.

A tank that is balanced overall, but has a low ceiling or high floor should have a low (near zero) score, despite being biased to players of a certain skill level.  This is intentional.

 

So, I'm sharing these results with you, to see if you have any feedback (or if I have missed any tanks that you think should be there)

If anyone is interested in how I got these numbers, just ask, but I'm not going to bother going into detail if people are not interested ;)

 

UPDATE

Version 3 stats:

 

View PostBaldrickk, on 28 August 2017 - 12:53 AM, said:

Ok, the method hasn't changed, but I have re-collected all the data (and made 100% sure that it is all EU data this time)

 

Get your copy of the graphs and scores here

 
Spoiler

 

 


Edited by Baldrickk, 28 August 2017 - 01:00 AM.


xx984 #2 Posted 30 July 2017 - 06:37 PM

    Major

  • Player
  • 55039 battles
  • 2,520
  • [GO-IN] GO-IN
  • Member since:
    08-11-2013

View PostBaldrickk, on 30 July 2017 - 05:34 PM, 

New tank? :unsure:



brumbarr #3 Posted 30 July 2017 - 06:47 PM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012
One request, with this one number tanks designed for a low number of players ( very bad or very good ) are being wrongly represented. I wonder if its possible to show 3 numbers? One total, one integrated from 45 to 50 and one from 55 to 65?


The 50b for example  is a fine tank but high skill ceiling, this 3 number system woulf show that. Or would that defeat the point of the rating?

Also i do start wondering how accurate the curves are seeing the m4 sherman next to the defender...

HaZardeur #4 Posted 30 July 2017 - 06:53 PM

    Second Lieutenant

  • Beta Tester
  • 33473 battles
  • 1,186
  • Member since:
    08-14-2010

View PostBaldrickk, on 30 July 2017 - 06:34 PM, said:

 

Balance Numbers

 

 

 

 

But tbh I have no clue what these numbers represent :amazed:


Edited by HaZardeur, 30 July 2017 - 06:54 PM.


anonym_YNch2j0j5oJ9 #5 Posted 30 July 2017 - 06:55 PM

    Warrant Officer

  • Player
  • 0 battles
  • 689
  • Member since:
    09-24-2018

T71. 1357f. Defender. Maus. Type5. Most OP tier for tier
Forget ur ratings.
Some tanks are balanced but many players suck in it, doesnt mean it needs buff

 

And is6 above t54 and patriot? Bro your ratings are bull..


Edited by Press2ForSkill, 30 July 2017 - 06:59 PM.


brumbarr #6 Posted 30 July 2017 - 07:01 PM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012

View PostPress2ForSkill, on 30 July 2017 - 06:55 PM, said:

T71. 1357f. Defender. Maus. Type5. Most OP tier for tier
Forget ur ratings.
Some tanks are balanced but many players suck in it, doesnt mean it needs buff

 

And is6 above t54 and patriot? Bro your ratings are bull..

Its not HIS ratings, its data gathered by a website put into one number. Not pulled out of anyones [edited]. The only faults in it are biases that show up in the data.



brumbarr #7 Posted 30 July 2017 - 07:01 PM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012

View PostPress2ForSkill, on 30 July 2017 - 06:55 PM, said:

T71. 1357f. Defender. Maus. Type5. Most OP tier for tier
Forget ur ratings.
Some tanks are balanced but many players suck in it, doesnt mean it needs buff

 

And is6 above t54 and patriot? Bro your ratings are bull..

Its not HIS ratings, its data gathered by a website put into one number. Not pulled out of anyones [edited]. The only faults in it are biases that show up in the data.



xx984 #8 Posted 30 July 2017 - 07:10 PM

    Major

  • Player
  • 55039 battles
  • 2,520
  • [GO-IN] GO-IN
  • Member since:
    08-11-2013


Element6 #9 Posted 30 July 2017 - 07:17 PM

    Field Marshal

  • Player
  • 29305 battles
  • 10,394
  • [T-D-U] T-D-U
  • Member since:
    01-06-2013

There are quite a few tanks in positions you wouldn't think possible in relation to how people speak of them. Type 4 Heavy at a negative number, and Tortoise at almost zero, among others.

 

Interesting read.



Junglist_ #10 Posted 30 July 2017 - 07:23 PM

    Second Lieutenant

  • Player
  • 37139 battles
  • 1,348
  • Member since:
    06-17-2013
T95 confirmed OP?

FluffyRedFox #11 Posted 30 July 2017 - 07:25 PM

    General

  • Player
  • 22971 battles
  • 8,394
  • Member since:
    12-05-2012
T2 light more OP than Defender :trollface:

Sfinski #12 Posted 30 July 2017 - 07:25 PM

    Major

  • Player
  • 31728 battles
  • 2,639
  • [-PJ-] -PJ-
  • Member since:
    09-26-2013

Like the idea. Brumbarr has a point. Dividing them to three numbers would surely help some to see difference in performance in different skill levels.

 

View PostElement6_TheSprout, on 30 July 2017 - 07:17 PM, said:

There are quite a few tanks in positions you wouldn't think possible in relation to how people speak of them. Type 4 Heavy at a negative number, and Tortoise at almost zero, among others.

 

Interesting read.

 

Type 4 fe. isn't OP, it's broken. Knowing the difference helps to join the conversations on the balance of the game. 



brumbarr #13 Posted 30 July 2017 - 07:33 PM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012

View PostElement6_TheSprout, on 30 July 2017 - 07:17 PM, said:

There are quite a few tanks in positions you wouldn't think possible in relation to how people speak of them. Type 4 Heavy at a negative number, and Tortoise at almost zero, among others.

 

Interesting read.

Type4 isnt OP , just is a combination of bad game mechanics.

 

I would expect the toirtuse to be there tbh, has lot of hp and dpm, so in the current meta will do fine.



Tidal_Force #14 Posted 30 July 2017 - 07:34 PM

    Lieutenant General

  • Player
  • 16192 battles
  • 6,839
  • [-MM] -MM
  • Member since:
    08-29-2012

Care to share your code?

 

EDIT: Hmm. 2 3 4 among my questions:

 

Are you weighting by player population in WR brackets?

Why are you using best fit?

I don't think it's possible to fit any general curve here, so I'm not sure why use fit instead of just comparing histograms?

If you are just fitting line (I hope you are not) then what you are using as error estimation for your data points?


Edited by Tidal_Force, 30 July 2017 - 07:42 PM.


Element6 #15 Posted 30 July 2017 - 07:39 PM

    Field Marshal

  • Player
  • 29305 battles
  • 10,394
  • [T-D-U] T-D-U
  • Member since:
    01-06-2013

View PostSfinski, on 30 July 2017 - 07:25 PM, said:

Like the idea. Brumbarr has a point. Dividing them to three numbers would surely help some to see difference in performance in different skill levels.

 

 

Type 4 fe. isn't OP, it's broken. Knowing the difference helps to join the conversations on the balance of the game. 

I didn't speak of OP or broken, but rather about where the tanks land on the list, based on how people talk about them. Some say it is OP, some say it is broken, some say it is OP and broken.

But while we are at it, you might be able to shed some light;

I've never heard the T95 spoken about as broken, nor OP, just highly situational. The Maus is, at least from what I can gather from forumites in the know, extremely overbuffed, I.E OP as fvk. Do I lack in information here, or is there another explanation for the T95 being rated at +30 points compared to the Maus? Looking at WR graphs it seems the T95 is boosting player's WRS even more than the Maus, yet I seldom see any threads about it, demanding nerfs.



brumbarr #16 Posted 30 July 2017 - 07:42 PM

    Lieutenant General

  • Player
  • 38626 battles
  • 6,326
  • Member since:
    07-30-2012

View PostElement6_TheSprout, on 30 July 2017 - 07:39 PM, said:

I didn't speak of OP or broken, but rather about where the tanks land on the list, based on how people talk about them. Some say it is OP, some say it is broken, some say it is OP and broken.

But while we are at it, you might be able to shed some light;

I've never heard the T95 spoken about as broken, nor OP, just highly situational. The Maus is, at least from what I can gather from forumites in the know, extremely overbuffed, I.E OP as fvk. Do I lack in information here, or is there another explanation for the T95 being rated at +30 points compared to the Maus? Looking at WR graphs it seems the T95 is boosting player's WRS even more than the Maus, yet I seldom see any threads about it, demanding nerfs.

My guess is the WR curves have a fatal flaw which i was going to write something about. Its about recent wr which shows on the graph to overal wr which is the reference line. Plotting one in function of another actually is quite wrong and creates biases escpecially for tanks that arent played much.



Sfinski #17 Posted 30 July 2017 - 07:45 PM

    Major

  • Player
  • 31728 battles
  • 2,639
  • [-PJ-] -PJ-
  • Member since:
    09-26-2013

View PostElement6_TheSprout, on 30 July 2017 - 07:39 PM, said:

I didn't speak of OP or broken, but rather about where the tanks land on the list, based on how people talk about them. Some say it is OP, some say it is broken, some say it is OP and broken.

But while we are at it, you might be able to shed some light;

I've never heard the T95 spoken about as broken, nor OP, just highly situational. The Maus is, at least from what I can gather from forumites in the know, extremely overbuffed, I.E OP as fvk. Do I lack in information here, or is there another explanation for the T95 being rated at +30 points compared to the Maus? Looking at WR graphs it seems the T95 is boosting player's WRS even more than the Maus, yet I seldom see any threads about it, demanding nerfs.

It's a list of tanks by their OP value. So, you wondering some tanks position on OP scale, you are talking about them being OP. 

 

As we've been trough, Brumbarr mainly, one value isn't enough to determine that. 

 

// Ninja'deded


Edited by Sfinski, 30 July 2017 - 07:46 PM.


laulaur #18 Posted 30 July 2017 - 07:56 PM

    Warrant Officer

  • Player
  • 46687 battles
  • 986
  • [BLAUR] BLAUR
  • Member since:
    08-11-2011

I guess people are interested, would you care to share by what witchcraft do you got those numbers.

in my opinion they do not reflect really well how some tanks feel now.



Element6 #19 Posted 30 July 2017 - 07:59 PM

    Field Marshal

  • Player
  • 29305 battles
  • 10,394
  • [T-D-U] T-D-U
  • Member since:
    01-06-2013

View Postbrumbarr, on 30 July 2017 - 07:42 PM, said:

My guess is the WR curves have a fatal flaw which i was going to write something about. Its about recent wr which shows on the graph to overal wr which is the reference line. Plotting one in function of another actually is quite wrong and creates biases escpecially for tanks that arent played much.

I see. I have been a little reluctant to accept WR curves as proof of tanks OP/UPness, which is why I have been trying to be a little sarcastic when talking about the FV4202, seeing as that tank's curve looks ok but people talking about it as if it is useless. It's also why I wrote the post about WG possibly spending a lot of time balancing a tank, not because they look at WR gaphs, but rather because they are looking at how large groups of players do against a tank. I simply proposed that as a plausible possibility. I have also written about the Defender and how the OPness you see in the WR graphs as declined a little, supporting the idea that the OPness has in fact gone down as people's ability to handle them has gone up, compared to the period just after release when "nobody" had a clue how to fight them effectively.

 

That being said, Maus has been played 324k times compared to 265k times for the T95, over the course of the last four weeks. So decent samplesize for both, though I would guess that Maus might have a higher average player skill level since it's and end-piece and not a tank that grinds to a new tank, like the T95.

This is why I have wanted to look at WGs server statistics for a long time, because there are so many parameters that ideally should be looked at that we do not have access to, and I guess I am not alone there.



RamRaid90 #20 Posted 30 July 2017 - 08:11 PM

    Lieutenant General

  • Player
  • 21507 battles
  • 6,487
  • [D0NG] D0NG
  • Member since:
    12-14-2014
If defender isnt at the top of the list, it's wrong.




1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users