Jump to content
Korean Random
Praetor77

[WN6, WN7 рейтинг]

Recommended Posts

So how was the scale recalculated? Did you decide on 0-99 or 0-999?

 

Glad to see you chose WN over eff... :D

 

How are the color scales? I am 91, but blue instead of purple...

 

 

Also, I dont believe xWN needs any rescaling of the scores, simply translate to a 0-99 scale using the data you have...

Lastly, it would be nice to have a different scale for each server...

Edited by Praetor77

Share this post


Link to post

Short link
Share on other sites
Lastly, it would be nice to have a different scale for each server...
true.

Share this post


Link to post

Short link
Share on other sites

 

Прошу помочь с включением WN6, а еще лучше залейте свой конфиг.  

 

Share this post


Link to post

Short link
Share on other sites

Shurasvolga, I´m not sure what you want to do. You are already using WN6 by the look of that config file. If you want to use the new scale, change every instance of "wn" to "xwn".

 

Guys maybe a 0-999 scale is better? 0-99 gets confusing with winrate...

Share this post


Link to post

Short link
Share on other sites

I have been trying to understand the new XVM and I will read more here (english) - but wanted to show you something in my last match.

 

It just seems that a match like this should not result in a lower eff.

 

1053 hit damage + 2840 DUD

 

It seems the eff calulation should use some of the games xp calculations - something seems off!

post-10280-0-73676600-1361211699_thumb.jpg

post-10280-0-41297700-1361211783_thumb.png

Edited by oldsurferjoe

Share this post


Link to post

Short link
Share on other sites

@oldsurferjoe,

Unfortunately, DUD impossible to be considered in calculating the rankings.

average damage by IS-6: 1031

TOP damage by IS-6: 2190

Share this post


Link to post

Short link
Share on other sites

i hope that the original standard wn6 (scale of 4 digits) to be chosen

Share this post


Link to post

Short link
Share on other sites

No, we will use XVM scale by default. 4-digit rating is anachronism.

Edited by sirmax

Share this post


Link to post

Short link
Share on other sites

Guys I wanted to let you know we have made some changes to WN.

 

We noticed the sea of red usually occurs in low tier matches, which was due to low tier avg due to small number of games. We corrected the "sealclubber" or "pedotanker" penalty so it will not apply to new players anymore. We also apply a penalty to "spots" for players with avg tier 1-2, since pedotankers get MANY spots due to high view range of tanks compared to enemies when used with 100% crew and equipment.

 

This is the WN7 formula:

 

 

(1240-1040/(MIN(TIER,6))^0.164)*FRAGS
+DAMAGE*530/(184*e^(0.24*TIER)+130)
+SPOT*125*MIN(TIER, 3)/3
+MIN(DEF,2.2)*100
+((185/(0.17+e^((WINRATE-35)*-0.134)))-500)*0.45
-5 - MIN(TIER,5))*125] / [1 + e^( ( TIER - (GAMESPLAYED/220)^(3/TIER) )*1.5 )]

Edited by Praetor77
  • Upvote 1

Share this post


Link to post

Short link
Share on other sites

+DAMAGE*530/(184*e^(0.24*TIER)+130)

 

Hey. Have an idea how to improve the correction of damage, depending on the technique, in which the player plays more often. Average level rather badly considers that the technique at the same level can be very different even within the same class. For example, the average damage by Chaffee twice more than the T-50-2, although they are both light tank level 5. I suggest doing accounting on the basis of standard tables of middle and top-end damage on each tank:

http://code.google.com/p/wot-xvm/source/browse/trunk/src/xvm/src/wot/utils/VehicleInfoData3.as

For example, a player had 2 fights at ltraktor and 3 fights on maus. We know the average damage on a server on these tanks: 83, 1736. We know the top-end damage on a server on these tanks: 377 and 2689. Then the average player should have an average damage (2 * 83 +3 * 1736) / (2 +3) = 1075 and a top player (2 * 377 +3 * 2689) / (2 +3) = 1764. And give score to depend on how the average real damage is related to these numbers.

And the same can be done with frags.

Edited by seriych

Share this post


Link to post

Short link
Share on other sites

A tank-specific rating should be more accurate than a global stat based one like WN7. The problem is the NA API gives some wrong stats with certain tanks. If that worked, then it would be interesting to devise a tank-specific rating, and what you describe sounds like a very good start.

Edited by Praetor77

Share this post


Link to post

Short link
Share on other sites

The problem is the NA API gives some wrong stats with certain tanks.

 I know that API is not true given the damage by tank. But the number of fights is given correctly, and that's all we need, apart from the reference table. Or NA API not give fights too? But if it is so, how do you calculate average level?

Edited by seriych

Share this post


Link to post

Short link
Share on other sites

P.S.I apologize for mistake. Analisys shoud be between WN7(WN6) vs. newEFF

 

7cc3f9ecf5956d4c.jpg

 

 

 

 

 

New "efficiency rating" formula:
DAMAGE * (10 / (TIER + 2)) * (0.21 + 3*TIER / 100)
FRAGS * 250 +
SPOT * 150 +
log(CAP + 1,1.732) +
DEF * 150;

 

WN7 formula:
(1240-1040/(MIN(TIER,6))^0.164)*FRAGS
+DAMAGE*530/(184*e^(0.24*TIER)+130)
+SPOT*125*MIN(TIER, 3)/3
+MIN(DEF,2.2)*100
+((185/(0.17+e^((WINRATE-35)*-0.134)))-500)*0.45
-5 - MIN(TIER,5))*125] / [1 + EXP( ( TIER - (GAMESPLAYED/220)^(3/TIER) )*1.5 )]

 

My conclusions to WN7(WN6) vs. newEFF  (DMG part formula):

* changes dependent on the DMG rebalanced from  tiers (WN7(WN6) vs. newEFF)
* bonus part of WN7 formula (DMG part)  for tiers: 1 to 7 ... higher tiers which generate more dmg --> value of part of WN7 formula (DMG part) is markedly cuted vs. newEFF

Edited by MaverS_71

Share this post


Link to post

Short link
Share on other sites

O_o

The damage part of the formula did not change...

 

Only the spots and pedotanker penalty changed. 



Now that I reread what you posted Seyrich, yeah you are right, we would only need the number of games played on each tank...interesting idea. If we simply use the topend damage/kill as 100% of score for each tank, that could be easily done... and sounds interesting! I wouldn´t use averages since they are very misleading. Appropriate multipliers would have to be devised for each term to make that into a proper scale.

 

I thought you were proposing to use stats for each tank, calculating a rating for each tank using top-end server stats for each tank, and then calculating the player´s score combining his score for each tank...

Share this post


Link to post

Short link
Share on other sites

O_o

The damage part of the formula did not change...

 

Only the spots and pedotanker penalty changed. 

 

Now that I reread what you posted Seyrich, yeah you are right, we would only need the number of games played on each tank...interesting idea. If we simply use the topend damage/kill as 100% of score for each tank, that could be easily done... and sounds interesting! I wouldn´t use averages since they are very misleading. Appropriate multipliers would have to be devised for each term to make that into a proper scale.

 

I thought you were proposing to use stats for each tank, calculating a rating for each tank using top-end server stats for each tank, and then calculating the player´s score combining his score for each tank...

It would be great if you realize this idea. You will be able to balance, for example, a low damage of LT and to refuse such disputable and hardly balanced parameter as "spot" which was entered into a formula most likely just as compensation of a low damage of LT. After all it is difficult to argue with that fact that in the itself "spot" parameter has no relation to a skill.

Edited by kkonstt

Share this post


Link to post

Short link
Share on other sites

I wouldn´t use averages since they are very misleading.

 No, I think we must take into account the average damage. Because the difference between the middle and top in different tanks is different. I think, give points need to something like this:

post-8781-0-58492700-1361970259_thumb.png

large points is the middle and top-end damage.

Share this post


Link to post

Short link
Share on other sites

In my humble opinion average damage is not good because it is VERY easy to get more than 80 damage in tier 1, but not so easy to get over 1730 dmg in tier 10...

 

Average damages get distorted because MANY more players play low tiers than high tiers, and also much better players play high tiers so... I would use top-seed to balance between tanks, you divide the players by the topseed damage or kills,  then you can multiply by whatever value you wish in order to maintain a similar balance to WN/efficiency scores.

Edited by Praetor77

Share this post


Link to post

Short link
Share on other sites

In my humble opinion average damage is not good because it is VERY easy to get more than 80 damage in tier 1

Example for MS-1:

post-8781-0-39296100-1361984364_thumb.png

as you see, to get a really high score, it is necessary to cause damage far above average.

 

but not so easy to get over 1730 dmg in tier 10...
We have statistics that shows the level of the average player on the tank. Why are these fantasies?

 

Average damages get distorted because MANY more players play low tiers than high tiers, and also much better players play high tiers so...
 This is already included in this distribution of points. And also considered that on different tanks to play above average harder or easier, which we can not consider, if we take only the top-end damage.

Share this post


Link to post

Short link
Share on other sites
Guest
This topic is now closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...