Jump to content
Garbad

Wn8 is deeply flawed.

  

166 members have voted

  1. 1. Wn8?

    • ...stop hating you shitlord DPG stat denying retard.
    • ...DPG is life, Wn62A is king.
    • ...personally, I think sniping in my light tanks is skilled gameplay.
    • ...Efficiency = pad tier 2, Wn4 = pad tier 5, wn7 = pad t10 td, wn8 = pad t10 med. Does this mean wn9 = pad in my Fv215?


Recommended Posts

I've always had my beef with metrics in general (imo, any focus on anything other than win rate is harmful), but backed down because Wn8 has actual statistical validity and methods.  It has value, especially as a tool for XVM to ballpark player competence.  Wn8 creators say that Wn8 is not designed as a recruitment, comparison, or Epeen tool...yet we all know, that's not the primary use of Wn8.  Instead, people use it to compare each other, to judge recruits, and to track progress. And when it comes to those uses, Wn8 just isn't effective.  Its heavily biased by tank, tier, and role selection.  I've written at length about this.  Worse, its extreme emphasis on DPG makes some players pad DPG at the expense of winning.  Yes, its ballpark valid on a population level, but for how its actually used, it just plain sucks.

Don't believe me?

Let's check the stats of recent, top Wn8 players.  Before I check, I'll tell you what we will see -- we will see massive amounts of games played in the T-62A/140, E-50/E-50M, and possibly the STB and leopard1.  Why?  Because these tanks have low expected values but high performance and are never on bottom.  This means players can use these scouty mediums as primary damage dealers, average 3-3.5k dpg, and attain extraordinary Wn8 scores...without having extraordinary win rates (3-3.5k dpg as a tier 10 isn't that amazing from a team carry perspective.  A blue TD should put that much out sniping side by side with a wn8 medium).  So I predict we see massive amounts of t10 medium farming with mediocre win rates.

Here are the top WN8 players in the last 30 days, 250 game minimum, 4k Wn8 minimum.
http://www.noobmeter.com/serverTop/na/RecentWN8/

#1 - Yato:
http://www.noobmeter.com/recent/na/YatogamiTohka/1004680695/1004680695_00012642_20140519_035552/1004680695_00013039_20140617_032909
396 of 397 games played in the T-62A

#2 - Evroz621
http://www.noobmeter.com/recent/na/Evroz621/1001026761/1001026761_00009329_20140518_081459/1001026761_00009612_20140613_142252
Top 2 most played tanks E-50 and E-50M

#3 - Nas
http://www.noobmeter.com/recent/na/Nas_/1000330864/1000330864_00019182_20140529_154542/1000330864_00019360_20140616_125300
Top 3 played tanks:  T-62A, E-50M, 140

#4 - E36M6
http://www.noobmeter.com/recent/na/E63M6/1001897109/1001897109_00017743_20140516_013102/1001897109_00018104_20140617_131530
Top 2 most played tanks:  E-50, E-50M

#5 - gotchamaster
http://www.noobmeter.com/recent/na/GotchMaster/1001376675/1001376675_00019908_20140524_062914/1001376675_00020226_20140617_080155
Most played tanks:  140, E-50M

The rest of the top ~25 looks similar:
http://www.noobmeter.com/recent/na/OIEIOIOF7I/1000274484/1000274484_00025556_20140518_174910/1000274484_00025782_20140617_052509
http://www.noobmeter.com/recent/na/downtown_girl/1001385041/1001385041_00024980_20140518_023327/1001385041_00025400_20140617_165312
http://www.noobmeter.com/recent/na/__ProMetheus__/1001638452/1001638452_00018003_20140515_073835/1001638452_00018268_20140617_172023
http://www.noobmeter.com/recent/na/MAMAsaidKNOCKuOUT/1004103560/1004103560_00019159_20140524_200125/1004103560_00019592_20140616_205624
http://www.noobmeter.com/recent/na/Nitts/1004090574/1004090574_00012660_20140516_071012/1004090574_00012902_20140617_021219
http://www.noobmeter.com/recent/na/PublicEn3my/1001643659/1001643659_00033425_20140525_035238/1001643659_00034465_20140617_172037
http://www.noobmeter.com/recent/na/gumiyaro/1003872009/1003872009_00017725_20140526_015543/1003872009_00018460_20140617_045806
http://www.noobmeter.com/recent/na/SmyleeRage/1003482352/1003482352_00013430_20140525_173349/1003482352_00013887_20140615_202223
http://www.noobmeter.com/recent/na/Russian_Pepper/1000789514/1000789514_00015441_20140523_221519/1000789514_00016048_20140616_142836
http://www.noobmeter.com/recent/na/Zrky/1004807434/1004807434_00004821_20140521_051526/1004807434_00005191_20140617_170152
http://www.noobmeter.com/recent/na/MichaelZordan/1000376820/1000376820_00019982_20140518_040016/1000376820_00020284_20140615_223431
http://www.noobmeter.com/recent/na/lord_recneps/1003605982/1003605982_00010068_20140513_064610/1003605982_00010426_20140617_172002
http://www.noobmeter.com/recent/na/Flashram/1001265192/1001265192_00016056_20140523_053104/1001265192_00016470_20140616_100603
http://www.noobmeter.com/recent/na/Dodoma/1000185415/1000185415_00028991_20140520_212839/1000185415_00030309_20140617_161541
http://www.noobmeter.com/recent/na/LanterneRouge/1004271012/1004271012_00020701_20140524_063719/1004271012_00021568_20140617_142840
^ a perfect 5 of his top 5 tanks are my 5 padding kings.

In fact, the only people who DIDN'T farm one of those three tanks yet got over 4k WN8 were die_weasel_mongol, Overlord_Prime, Shrewmz, "_Intrepid," AutismSpeaks (lol, elc), "Diastant," and CanadianImpact.  You will note that virtually all of these players play almost all games in their chosen padder, and most of them did not play that tank a few months ago (before Wn8 came out).  This suggests that players are not running that tank because they enjoy it or the meta, but instead because they learned it can inflate their Wn8.  Its also worth pointing out that many (but not all) of these players have lack luster win rates, ranging from 59% to about 66%.  This suggests that putting out 3-3.5k DPG in a t10 medium, while impressive, is not particularly useful at winning games.  In addition, most of these players are run of the mill purples -- good players, but not the best players on the server.  Of the top 50 Wn8 players, probably only Carbonward, Yato, and Canadian would be considered a consensus top 25 player by their peers (maybe a few others).  So in essence, EVERY top Wn8 player is deliberately stat padding.  Much like people playing tier 2 to farm Efficiency, many players have clearly figured out how to pad Wn8.  Like efficiency, people aren't playing to win more, to get better, or because they enjoy clubbing seals...they are running what the metric rewards to make themselves look better than they are, at the expense of winning.  And that's just plain degenerate.

So with that in mind, here is what needs to be done:

1.  Reduce Wn8's weight on DPG overall.  While the correleation is great low end and with heavy/TDs, it breaks down high end and with lights/meds.
2.  Implement a two point system for DPG.  One point is average, second is top 1% DPG.  This means tank selection wouldn't matter.
3.  Fix the class/tank/tier balance.  If 4k Wn8 is the peak in some tanks, similar scores should be attainable in all tanks.
4.  Increase the expected value for the T-62A, Object 140, E-50, E-50M, and STB.  Follow the ensuing tears to their source, and you will know who was padding all along.
5.  Triple the ELC expected value to make autismspeaks shit a brick.

And here we go.

Link to post
Share on other sites

5.  Triple the ELC expected value to make autismspeaks shit a brick.

I LOL'd IRL.

 

I actually do agree with you, Garbad, but I'm not sure at this point what can *truly* be done to fix it since at the end of the day people will game whatever system is created in order to pad their egos and stroke their epeens.

Link to post
Share on other sites

I remember a thread here on wotlabs some time ago named "Is playing tanks you find fun statpadding eventhough they are t10s" or something like that. Wellp. Statpadding can be/is done even in T10s. Now to all those who will come at me saying i'm wrong please tell me more how it's coincidence that all of you have the same favourite tanks *cough* E50s-russianstronkmeds *cough* Pretty much any measurment tool(eff,WNs) have flaws and sooner or later they are abused and padded heavily. It just takes time for players to find perfect padding combo.

Link to post
Share on other sites

I remember a thread here on wotlabs some time ago named "Is playing tanks you find fun statpadding eventhough they are t10s" or something like that. Wellp. Statpadding can be/is done even in T10s. Now to all those who will come at me saying i'm wrong please tell me more how it's coincidence that all of you have the same favourite tanks *cough* E50s-russianstronkmeds *cough* Pretty much any measurment tool(eff,WNs) have flaws and sooner or later they are abused and padded heavily. It just takes time for players to find perfect padding combo.

To be fair, being OP is fun.  Pretty much no one says the v39 is their favorite tank, we all gravitate to the waffles instead.

Link to post
Share on other sites

To be fair, being OP is fun.  Pretty much no one says the v39 is their favorite tank, we all gravitate to the waffles instead.

Why are you opening up old wounds!? How did you knew i hate V39 more than Lee?! Also waffles suck at WN8. But everyone loves dmg padding. Am i rite?

Link to post
Share on other sites

The one thing I like about Wn8 is it is deterring the best players from playing the most OP tanks atm.  The fewer cums we have in waffles/hellcats/etc the better for everyone else, although it makes the blues with 4k dpg in their waffle all the more glaring, because we know the damage padders would be shitting out more, except for wn8 penalty.

Link to post
Share on other sites

I am 50/50 on whether or not I like WN8. I definitely agree with the problems you've outlined with it. As a light tank/medium tank driver, I tend to #YOLO more often than is probably necessary. If I see a good chance to break a flank by doing something stupid and I feel like my team can take advantage of it, I sure as hell am going to do it.

 

I think that WN8 rewards a passive, damage-farmy playstyle. I've gotten some ridiculously high WN8 games by playing like a base camping TD in tanks that should be on the front lines. Oneeechan demonstrated just how effective passive play can be at padding WN8 with his ELC padding challenge.

 

However, as I get better at staying alive while #YOLOing, my DPG goes up along with my WN8. In that regard, I think WN8 accurately reflects my improvement in skill.

 

You can certainly game WN8 but I still think it's a pretty sound way of getting a feel for a player at a glance. As always, platooning with a player is the best way to evaluate their skill and there are no shortcuts there.

Link to post
Share on other sites

So with that in mind, here is what needs to be done:

1.  Reduce Wn8's weight on DPG overall.  While the correleation is great low end and with heavy/TDs, it breaks down high end and with lights/meds.

2.  Implement a two point system for DPG.  One point is average, second is top 1% DPG.  This means tank selection wouldn't matter.

3.  Fix the class/tank/tier balance.  If 4k Wn8 is the peak in some tanks, similar scores should be attainable in all tanks.

4.  Increase the expected value for the T-62A, Object 140, E-50, E-50M, and STB.  Follow the ensuing tears to their source, and you will know who was padding all along.

5.  Triple the ELC expected value to make autismspeaks shit a brick.

And here we go.

 

  1. Badguy, VBE is slowly approaching. Since WN7's weight on frags/other things was higher than WN8's, I expect the weight of DPG to decrease in it and other stats like frags increase, but I think DPG will still be king. I may actually limit the value of each stat using all of the other stats, instead of hinging the entire rating on damage, because of this issue.

     

  2. Nope. Average/top 1% wouldn't work geometrically. Low skill ceiling tanks would be easy WN8 (WN9) padders for above average players, and they would get totally screwed below average. It's better to find relative variance between tanks at approximately the average range.

     

  3. Some tanks don't perform much better with skill at a certain point, so you'd be giving free points to low skill ceiling tanks. Refer to #2.

     

  4. The next expected values update is likely going to help with TX MT padding, as it seems to line up each tank in a more logical way. WN8 is still WN8 though, maybe WN9 will have less weight on it.

     

  5. Will probably happen, even if we have to heavily screw with the numbers.

Link to post
Share on other sites

Voted shitlord option for lack of bacon. As far as the E-50 goes, it's currently my only tier 9 and soon to be my first and only tier 10. Here I was convincing myself my recent stat improvements were finally getting used to (and better) at higher tier play as well as stat-padding. Didn't realize the E-50 was that paddable.

Link to post
Share on other sites

As soon as they announced WN8 I realized that the WN team doesn't "get it". If I'm not mistaken, one of the reasons for WN8 superseding WN7 was to close "padding" loop holes in WN7. What people don't understand is that metrics CAUSE padding. You start measuring something, someone will figure out a way to game the system. It's human nature. Stop worrying about it.

 

For example, if you over-perform in a tier 4, why is that not as valued as over-performing to the same degree in a tier 10? Because someone decided that was "seal clubbing" and seal clubbing is bad. I'm not advocating seal clubbing, I find tier <4 too boring to contemplate any serious investment, but metrics shouldn't be geared to "proper" behaviour. They should just report data. 

 

It would have happened eventually but WN8 has made me check out of the whole stats thing. I pretty much just focus on wins now.

Link to post
Share on other sites

I felt this was a major issue with WN8 when it was being created.  It was sold as something that would allow you to play what you'd like rather then avoid underpowered tanks.  To some extent that has worked but if you are trying to improve your WN8 as many of us on here are doing then really you have to play certain under valued tanks.  It seems to favor a deep garage.

 

The issue with switching things up is that when tanks or meta changes do you base WN8 off the old stats or the new ones?

Link to post
Share on other sites

I will fully admit that I've been playing my Tier 10 mediums a lot more than heavies or lower tier tanks.

I simply find them the best for winning matches, coincidentally they're also the best for padding WN8. And although it doesn't really mean anything, breaking 4k has just been a little personal goal for me.

I do agree that 4k should be attainable in all tanks and not just a select few mediums.

Link to post
Share on other sites

I will fully admit that I've been playing my Tier 10 mediums a lot more than heavies or lower tier tanks.

I simply find them the best for winning matches, coincidentally they're also the best for padding WN8. And although it doesn't really mean anything, breaking 4k has just been a little personal goal for me.

I do agree that 4k should be attainable in all tanks and not just a select few mediums.

Yeah, I can't blame you for that.  I enjoy meds as well.  Plus there was the STB event recently.  Its not as noxious padding as the t2 abusers for sure, but its still padding.

Link to post
Share on other sites

You know, come to think of it, I thought it was kind of strange that so many people voted flex mediums as the best solopub tanks for winning in I am really of the opinion that beefy, armored tanks (Brutes & Heaviums) are the best to secure a win for average-to-good players based on my own observations. I can hit high 60s solopub winrate in tanks like the IS-6 but average only high blue WN8 in them.
 
Now it's starting to click that people are probably voting based on which tank subclass gets them decent winrate but ridiculous WN8.

Link to post
Share on other sites

You know, come to think of it, I thought it was kind of strange that so many people voted flex mediums as the best solopub tanks for winning in I am really of the opinion that beefy, armored tanks (Brutes & Heaviums) are the best to secure a win based on my own observations. I can hit high 60s solopub winrate in tanks like the IS-6 but average only high blue WN8 in them.

 

Now it's starting to click that people are probably voting based on which tank subclass gets them decent winrate but ridiculous WN8.

I noted that as well.  Although in tier 10 mediums are good at solopub, better than heavies.  But in mid/low tiers heavies are clearly king.

Link to post
Share on other sites

Everything is muddled. 

 

Also, there are plenty of damage farmers who have close or 70% wr solo. Aren't you getting that 70% tier 10 wr from your platoons garbad? Also 4400 wn8 isn't impressive in the t10 mediums. The real top players have over 5000.

Link to post
Share on other sites

5.  Triple the ELC expected value to make autismspeaks shit a brick.

 

All of this

 

and If I was in that top 50 Wn8, I was padding in the T29(recent tourney) and pershing (Need 4th skill nao)

Expect me to drop off and steady out at like 3700-3800 or lower shortly (grinding crews on low tier tourney tanks)

Link to post
Share on other sites

Everything is muddled. 

 

Also, there are plenty of damage farmers who have close or 70% wr solo. Aren't you getting that 70% tier 10 wr from your platoons garbad? Also 4400 wn8 isn't impressive in the t10 mediums. The real top players have over 5000.

I can't decide if this poast is serious.

Link to post
Share on other sites

Well, most easy to pad tanks tend to be flexible but with no obvious exploitable advantages, pubbies tend to suck in these as opposed to pros who value the flexibility.

 

Which is why most pros loath TDs, not that its shit for wn8, its just shit for winning due to its inflexibility.

 

Then again, theres a wn8 pad tank in just about every tier

 

Tier 4: T-28

Tier 5: ELC

Tier 6: 56-16/12t

Tier 7: t20, tiger, kv13

Tier 8: Pershing, 416

Tier 9: E50

Tier 10: Is7, E100, all meds.

Link to post
Share on other sites

See, now this is why I like Garbad's poasting. I've been struggling to break the 3k WN8 mark, and was wondering what I was doing wrong. Spent a solid 2 hrs going over my latest replays looking for flaws in my play and whatnot. Then it hit me, my gameplay has been solid (with those occassional tater games/suicidal pubbie teams), I've been playing in all the "wrong" tanks. Whenever I go on a Leo 1 spree, my WN8 skyrockets. Then I go back to grinding tiers 5-9, and it drops. The whole getting bottom-tier shafted and playing mediocre tanks chews up and spits out my WN8, which doesn't necessarily mean I all of a sudden turned into a potato. This poast kinda affirms my own suspicions. Thanks dude!

 

(P.S. I know last night I played my T-54 almost exclusively, I'm trying to get the 140 so I can build a clam wars inventory. E-100 and Leo 1 don't exactly cut it)

Link to post
Share on other sites

Totally agree. But to play devil's advocate on one smallish point:

 

What about people who pad WR through platoons? You can increase your WR without becoming a better player simply by platooning, right? Assuming you're not platooning with a 44%'r. You can increase WR by platooning with someone who is better than server average, because the average value of a player on a team is, by definition server average. You don't even have to communicate with them, just ensuring that they're there, instead of a 48%'r (server average).

 

Enter the third party performance metric, who's purpose is to measure individual performance, unaffected by platooning. Only problem is WN8 happens to measure DPG, and that inconsistently across different tanks. What we really want is to measure a player's contribution to winning the game, which is much more than DPG. 

 

 

This is where I'd promote the idea of a ranked elo based solo queue ladder. You Queue up for a game, and play ranked games against other players with similar elo's (who also elected to play ranked instead of normal pub games). You win, you gain elo, you lose, you lose it. Eventually you'll hit point where you're recent WR will flatten out to ~50%, and bam, that's how good a player you are.

Link to post
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...