Jump to content


Close
Photo

WN8


  • Please log in to reply
483 replies to this topic

#1 Praetor77

Praetor77

    IOC Applicant

  • The Formulaetor
  • 1,578 posts
  • Server: NA

Tanker Stats
Praetor77 [BADGR]

59.97%
29716
7.3

Posted 10 July 2013 - 02:40 PM

Hi guys, I still don't have much free time on my hands to do any work, but I thought we could start discussing WN8.

 

For those of you who follow my WN thread on the off forums, the main proposals for WN8 are:

 

*Eliminate tier based penaltys to frags and spots

*Half the contribution of defense points to WN, and lower the cap from 2.2 to 1.4.

*Possible re-work of damage curve (I would retailor the curve to values for each tier using top100 damage for each tank available using XVM and in-game stats)

*Re-implement the low tier penalty due to it being easily evaded by playing equal amounts of high and low tier games. Replace it by a NORMALIZER term, intended to compensate for sealbashing in tiers 1-3, which is calculated based on the number of games played on the 3-4 most played tanks of tiers 1-3. The idea is to have a cutoff which allows "normal" players to play many different low tier tanks without being penalized, while penalizing low tier sealbashers, INDEPENDENTLY of their avg tier. The current penalty in WN7 can be evaded if you raise your avg tier above 4, no matter how many games you spent sealbashing in tier 1.

 

 

Ideas? Commentarys? 

 

PS: I didnt want to make a mile-long post bringing all the stuff from the other thread, you can go check it to see details if you wish, but its full of stat-denying annoying little kids asking for a penalizer to players who play tier 10s o_o.

 

You can look at pages 100 and onwards, but this post pretty much explains everything>

 

http://forum.worldof...16#entry4343616


  • 9

#2 Tedster59

Tedster59

    Shoots The Plane on Airfield

  • Verified Tanker [NA]
  • 764 posts
  • Server: NA
  • Locationgrinding crew skills solopubbing top tier arty

Tanker Stats
Tedster59 [FOXEY]

53.3%
26329
7.29

Posted 10 July 2013 - 02:41 PM

sounds good.


  • 0
Stats Improvement in progress.
Posted Image

#3 CraBeatOff

CraBeatOff

    Tryhard Pedant

  • Head Moderator
  • 2,503 posts
  • Server: NA
  • LocationThe Seafood Buffet

Tanker Stats
CrabEatOff [PBKAC]

64.82%
12251
6.87

Posted 10 July 2013 - 02:43 PM

Is it feasible to run an analysis to see if confederate medals are in any way correlated? The conditions of the confederate are not easily manipulated, and generally good gameplay is rewarded by this medal. It also will have no overlap with kills/game. 


  • 0
 

Arty and TDs both promote(d) the same meta. I hate that meta.

 

 


#4 Marduk

Marduk

    Stats not good enough, deported to Mexico

  • Verified Tanker [NA]
  • 1,589 posts
  • Server: NA
  • LocationColorado

Tanker Stats
Marduk10 [PBKAC]

55.95%
18799
6.5

Posted 10 July 2013 - 02:46 PM

I suspect its going to be too rare an event Crab. Its why hockey stats are moving away from goals towards shots on net. Goals are random things that rarely happen.


  • 0
Posted Image

#5 Tedster59

Tedster59

    Shoots The Plane on Airfield

  • Verified Tanker [NA]
  • 764 posts
  • Server: NA
  • Locationgrinding crew skills solopubbing top tier arty

Tanker Stats
Tedster59 [FOXEY]

53.3%
26329
7.29

Posted 10 July 2013 - 02:49 PM

Is it feasible to run an analysis to see if confederate medals are in any way correlated? The conditions of the confederate are not easily manipulated, and generally good gameplay is rewarded by this medal. It also will have no overlap with kills/game. 

 

I have a feeling that it would be correlated to the alpha of tanks.  high rof plinkers are likely to plink a bunch of different tanks, while big alpha tanks likely will likely kill the tanks.


  • 1
Stats Improvement in progress.
Posted Image

#6 KraftLawrence

KraftLawrence

    IOC Applicant

  • Verified Tanker [NA]
  • 1,502 posts
  • Server: NA

Tanker Stats
KraftLawrence [FOXEY]

62.28%
16334
6.91

Posted 10 July 2013 - 02:50 PM

*Re-implement the low tier penalty due to it being easily evaded by playing equal amounts of high and low tier games. Replace it by a NORMALIZER term, intended to compensate for sealbashing in tiers 1-3, which is calculated based on the number of games played on the 3-4 most played tanks of tiers 1-3. The idea is to have a cutoff which allows "normal" players to play many different low tier tanks without being penalized, while penalizing low tier sealbashers, INDEPENDENTLY of their avg tier. The current penalty in WN7 can be evaded if you raise your avg tier above 4, no matter how many games you spent sealbashing in tier 1.

Can you explain this normalizer term a bit more? The idea sounds really cool btw :)


  • 0

#7 Tedster59

Tedster59

    Shoots The Plane on Airfield

  • Verified Tanker [NA]
  • 764 posts
  • Server: NA
  • Locationgrinding crew skills solopubbing top tier arty

Tanker Stats
Tedster59 [FOXEY]

53.3%
26329
7.29

Posted 10 July 2013 - 02:55 PM

Can you explain this normalizer term a bit more? The idea sounds really cool btw :)

 

sounds to me to be something like (battles in top 3 low tier tanks [tier < n])/(total battles)

 

interesting concept.  in this formula, the higher the number, the more sealclubbing.  n can be debated as there are different arguments for sealclubbing tiers.


  • 0
Stats Improvement in progress.
Posted Image

#8 Praetor77

Praetor77

    IOC Applicant

  • The Formulaetor
  • 1,578 posts
  • Server: NA

Tanker Stats
Praetor77 [BADGR]

59.97%
29716
7.3

Posted 10 July 2013 - 02:56 PM

Crab, I have no way to test such a thing, but it would sure be interesting. However, Killtech did a correlation using medals, and got some interesting, albeit difficult to interpret, results.

 

http://forum.worldof...47#entry4048647

 

http://forum.worldof...85#entry4213785

 

 

 

PS> You guys are posting faster than I can answer! XD

 

PS2> The normalizer term would be something like (%gamesin3toptier1-3 - cutoff% according to games played)*26.

Cutoffs would be something like>

 

 

At 0 games, the cutoff top 3 tier 1-3 tank should be 100%. At 750 games, the cutoff should be 45%. At 1500 games, the cutoff should be 23%. At 2500 games, the cutoff should be 15%. At 4000 games, the cutoff should be 11%. At 7500 games, cutoff should be 7.5%. At 12500 games, cutoff should be 5%. At 17500 games and higher, cutoff should be 4%. At 20000 or more games,cutoff should be 3.5%.

 


  • 0

#9 CraBeatOff

CraBeatOff

    Tryhard Pedant

  • Head Moderator
  • 2,503 posts
  • Server: NA
  • LocationThe Seafood Buffet

Tanker Stats
CrabEatOff [PBKAC]

64.82%
12251
6.87

Posted 10 July 2013 - 03:00 PM

I suspect its going to be too rare an event Crab. Its why hockey stats are moving away from goals towards shots on net. Goals are random things that rarely happen.

 

You're obviously just jealous of my 2.4% confed/game, compared to your 0.8% confed/game. :-D 

 

I really do think its a telling stat though. Its means you contributed to at least 6 kills for your team, right? And to do so implies survival, gun activity, effectiveness of your tank (not just last standing a kill). I realize this is just a construct, but I think that the medal really does imply superior gameplay. And its mutually exclusive with kills/game. And its really hard to farm/impossible?

 

However its rarity will of course FUCK PEOPLE'S SHIT when they look at their game by game stats and short term performances and it is included. I can already imagine the bitching "give damage more weight, confederate has too much weight, see this game has 4000 WN7 just because I got confederate"


  • 0
 

Arty and TDs both promote(d) the same meta. I hate that meta.

 

 


#10 CraBeatOff

CraBeatOff

    Tryhard Pedant

  • Head Moderator
  • 2,503 posts
  • Server: NA
  • LocationThe Seafood Buffet

Tanker Stats
CrabEatOff [PBKAC]

64.82%
12251
6.87

Posted 10 July 2013 - 03:07 PM

I have a feeling that it would be correlated to the alpha of tanks.  high rof plinkers are likely to plink a bunch of different tanks, while big alpha tanks likely will likely kill the tanks.

 

Right now high kill tanks are the alpha and autoloader ones. Any spike will do. Adding a little balance to that could be a good thing? I dunno, when I think about the stuff WN7 looks at right now, I'm just trying to find something that catches the "otherness" of quality play. Experience would of course be the best...


  • 0
 

Arty and TDs both promote(d) the same meta. I hate that meta.

 

 


#11 Praetor77

Praetor77

    IOC Applicant

  • The Formulaetor
  • 1,578 posts
  • Server: NA

Tanker Stats
Praetor77 [BADGR]

59.97%
29716
7.3

Posted 10 July 2013 - 03:12 PM

I later perfected the normalizer term to use sumof3tiers. This means the cutoff is based on the tier of your 3 most played tier 1-3 tanks. So, penalty starts earlier if your most played tier 1-3 tanks are all tier 1 tanks, and later if they are all tier 3 tanks. Does that make sense?

 

So...

 

summing the tiers of the three tanks and multiplying that by a factor, so if your top 3 played tier 1-3 tanks are all tier 3, sumof3tiers = 9, if they are all tier 1 sumof3tiers =3. This way, we can more easily catch sealclubbers who play tier 1 and 2 a lot, and differentiate them more easily from non-prem players, who will normally have more games on higher tier tanks,aka tier 3, so maybe:

At 0 games, the cutoff top 3 tier 1-3 tank games should be 300 games.
At 750 total games played, the top 3 tier 1-3 tank games cutoff should be 37*sumof3tiers (330 games, mostplayed all tier 3)
At 1500 games, the top 3 tier 1-3 tank cutoff should be 38*sumof3tiers (345 games, mostplayed all tier 3).
At 2500 games, the top 3 tier 1-3 tank cutoff should be 42*sumof3tiers (375 games).
At 4000 games, the cutoff should be 47*sumof3tiers (425 games)
At 7500 games, top 3 tier 1-3 tank cutoff should be 61*sumof3tiers (550 games)
At 12500 games, top 3 tier 1-3 tank cutoff should be 70*sumof3tiers (625 games)
At 17500 games and higher, top 3 tier 1-3 tank cutoff should be 78*sumof3tiers (700 games).
At 20000 or more games, top 3 tier 1-3 tank cutoff should be 79*sumof3tiers (710 games).
At 30000 or more games, top 3 tier 1-3 tank cutoff should be 81*sumof3tiers (730 games).

 

 

This is the final formula for the normalizer term I came up with. Someone with more programming skills than I could maybe give me a way to test it? Maybe we can test it on the wotlabs DB neverwish?

 

22 * (gamesintop3tier1-3 - sum3tiers * 83/(1+10^(0.15 – 0.000072*totalgames)) )

 

 

As for the damage curve do you guys think it needs a rework?


  • 0

#12 Marduk

Marduk

    Stats not good enough, deported to Mexico

  • Verified Tanker [NA]
  • 1,589 posts
  • Server: NA
  • LocationColorado

Tanker Stats
Marduk10 [PBKAC]

55.95%
18799
6.5

Posted 10 July 2013 - 03:13 PM

You're obviously just jealous of my 2.4% confed/game, compared to your 0.8% confed/game. :-D 

 

I really do think its a telling stat though. Its means you contributed to at least 6 kills for your team, right? And to do so implies survival, gun activity, effectiveness of your tank (not just last standing a kill). I realize this is just a construct, but I think that the medal really does imply superior gameplay. And its mutually exclusive with kills/game. And its really hard to farm/impossible?

 

However its rarity will of course FUCK PEOPLE'S SHIT when they look at their game by game stats and short term performances and it is included. I can already imagine the bitching "give damage more weight, confederate has too much weight, see this game has 4000 WN7 just because I got confederate"

 

LOL. Not really.

 

To me Confed means either the team had MAD focus fire and I was simply the fastest on the trigger or I kept having to snipe and back off because return fire was scary. I'm pretty convinced that staying on 1 tank till it is DEAD helps my team more than flinging out crap tons of damage at lots of tanks, but leaving 3 or 4 at 50 hp. Strangely enough, the analysis of Kills -> W/R vrs damge ->W/R agree with me. :)

 

Confed is NOT impossible to farm. I specifically attempt it during the "1000exp for confed" by switching targets. I mean, it doesn't always happen cause I don't PASS up a kill to drive elsewhere. Its harder to farm, I agree.

 

 

 

Again Crab, I suggest you look up a little at the "Fancy Stats" debate in Hockey. Specifically Corsi or Zone Entry. "Broad Street Bullies" blog had a GREAT analysis where they counted how often a player moved the puck from 1 zone to the other. They determined that zone entry correlated MUCH more strongly with wins than anything else. It was a surprise even for the guys doing the analysis. It showed that great players drive play to the enemy net. Goals are pretty random, but driving the puck into the enemy zone often results in good things. (enemy penalties, face offs in the offensive zone, shots on net) 

 

 

I don't think there should be any bonus for ANY of the medals. Not because they may or may not correlate... but because the EFFECT measuring will have on people chasing those medals. Look... Crucial Contrib and BIA are my favorite medals. They are just cool to me. 3 man BIA is pretty rare and shows you ALL did fairly well. (and lived) I don't think you should get a WN7 boost. Measure the common things. Leave the rare events as rare events that just happen sometimes.


  • 0
Posted Image

#13 Marduk

Marduk

    Stats not good enough, deported to Mexico

  • Verified Tanker [NA]
  • 1,589 posts
  • Server: NA
  • LocationColorado

Tanker Stats
Marduk10 [PBKAC]

55.95%
18799
6.5

Posted 10 July 2013 - 03:14 PM

I'll note I'm pretty happy with WN7 as it stands now. Its a pretty solid measure of how good someone is. At the purple/platinum level you have to dig deeper into the stats. Which is fine.


  • 0
Posted Image

#14 Praetor77

Praetor77

    IOC Applicant

  • The Formulaetor
  • 1,578 posts
  • Server: NA

Tanker Stats
Praetor77 [BADGR]

59.97%
29716
7.3

Posted 10 July 2013 - 03:25 PM

I agree it´s pretty decent, but there are problems with tier 1/10 farmers, and also many people who just like to play low tier tanks are currently recieving a huge penalizer that they dont deserve. I think the new normalizer term woudl solve both problems on the spot.

 

Also, the leader boards would make much more sense, wiping Ebichu, zealpath and others off the top 10.

 

Although, you could say othertanks are also sealclubbers, (m4 HEAT spam, 3601 konisch, T57 HEAT spam, T54HEAT, T69HEAT, etc.) but I would rather stay away from going down the per-tank road, as it is a fugly one, but stick with the low tier rule, since that is undisputable.


  • 0

#15 KraftLawrence

KraftLawrence

    IOC Applicant

  • Verified Tanker [NA]
  • 1,502 posts
  • Server: NA

Tanker Stats
KraftLawrence [FOXEY]

62.28%
16334
6.91

Posted 10 July 2013 - 03:29 PM

What I don't understand about WN7 is the sort of staticness of the penalizer towards tier 1-3. I don't feel that the skill level of WoT players is separated into such a neat set of 2 bins (seal-club tiers, and non-seal-club tiers).

 

What if there were something more fluid - you get more points for playing 6 as opposed to 5, you get more points for tier 9 as opposed to 8, etc.? I think this would be more representative than just "1-3, 4-10".


  • 0

#16 CraBeatOff

CraBeatOff

    Tryhard Pedant

  • Head Moderator
  • 2,503 posts
  • Server: NA
  • LocationThe Seafood Buffet

Tanker Stats
CrabEatOff [PBKAC]

64.82%
12251
6.87

Posted 10 July 2013 - 03:29 PM

I agree it´s pretty decent, but there are problems with tier 1/10 farmers, and also many people who just like to play low tier tanks are currently recieving a huge penalizer that they dont deserve. I think the new normalizer term woudl solve both problems on the spot.

 

Also, the leader boards would make much more sense, wiping Ebichu, zealpath and others off the top 10.

 

Although, you could say othertanks are also sealclubbers, (m4 HEAT spam, 3601 konisch, T57 HEAT spam, T54HEAT, T69HEAT, etc.) but I would rather stay away from going down the per-tank road, as it is a fugly one, but stick with the low tier rule, since that is undisputable.

 

Although, as a tier1/2 player I've got to say, that since new accounts get 100% crews with 8.6 and accuracy is MUCH better than before, and arty is now in nearly ever battle tier...low tier play is harder than it was prior to the patch.


  • 0
 

Arty and TDs both promote(d) the same meta. I hate that meta.

 

 


#17 Praetor77

Praetor77

    IOC Applicant

  • The Formulaetor
  • 1,578 posts
  • Server: NA

Tanker Stats
Praetor77 [BADGR]

59.97%
29716
7.3

Posted 10 July 2013 - 03:36 PM

@Kraft, that would only result in a stat that rewards higher tier play.

 

 

@Crab:

True, but there is always the problem of people with 5000 games in their MS-1 already.

 

Also, take into account this new normalizer term would only affect you if you play LOTS of games on the same top 3 tier 1-3 tanks.

 

 

 

It would be awesome if people could download this spreadsheet, fill it out with their overall stats, and games played on their top 3 tier 1-3 tanks, and see what their WN7 is with or without their top 3 played tier 1-3 tanks. post results, and that would help me calibrate WN8 normalizer term. This would only make sense if you have at least 2 tier 1-3 tanks with over 150 games.

 

http://www.mediafire...bc1d5njcswab1kj

 

 


  • 2

#18 CraBeatOff

CraBeatOff

    Tryhard Pedant

  • Head Moderator
  • 2,503 posts
  • Server: NA
  • LocationThe Seafood Buffet

Tanker Stats
CrabEatOff [PBKAC]

64.82%
12251
6.87

Posted 10 July 2013 - 03:42 PM

@Kraft, that would only result in a stat that rewards higher tier play.

 

 

@Crab:

True, but there is always the problem of people with 5000 games in their MS-1 already.

 

Also, take into account this new normalizer term would only affect you if you play LOTS of games on the same top 3 tier 1-3 tanks.

 

 

 

It would be awesome if people could download this spreadsheet, fill it out with their overall stats, and games played on their top 3 tier 1-3 tanks, and see what their WN7 is with or without their top 3 played tier 1-3 tanks. post results, and that would help me calibrate WN8 normalizer term. This would only make sense if you have at least 2 tier 1-3 tanks with over 150 games.

 

http://www.mediafire...bc1d5njcswab1kj

 

I've only got a single tier 1-3 tank over 100 games (201 in the Cruiser III). Does this disqualify me?


  • 0
 

Arty and TDs both promote(d) the same meta. I hate that meta.

 

 


#19 KraftLawrence

KraftLawrence

    IOC Applicant

  • Verified Tanker [NA]
  • 1,502 posts
  • Server: NA

Tanker Stats
KraftLawrence [FOXEY]

62.28%
16334
6.91

Posted 10 July 2013 - 03:47 PM

Is rewarding higher-tier play bad though?

From personal experience, I feel like I can carry on my T49 or hellcat really easily. But put me in a T-54 or Conqueror - ya I still can do over pretty decent DPG, but I can't carry the games nearly as well. It's way harder. People who can truly carry these tiers should be rewarded for carrying in a harder environment, I think.

 

What I'm saying is - currently with WN7, I could just inflate my stats by constantly playing tiers 5-6, where I get no penalty and get massive stats. There's no reason for me to lower my stats playing tier 10 where I can't carry. If WN8 gave more incentive to play higher tiers, it would be nice.


  • 1

#20 CrazyScientist

CrazyScientist

    Afraid To Solo

  • Verified Tanker [NA]
  • 605 posts
  • Server: NA

Tanker Stats
CrazyScientist [GMONK]

51.5%
15620
7.02

Posted 10 July 2013 - 03:48 PM

I do think the damage and kills need a re-balance.  (As stated before I have no proof--only that as I go up in tier, it feels it gets harder and harder to acheive "good games"--realizing of course that a game by game basis is pretty much the antithesis of stats but still--I HAV ETO WHINE ABOUT SOMETHING!)  Tier 9 and 10 are almost a different game than tier 6-7-8--in fact there are several sub-bands of WoT:  Tier 1-3, Tier 4/5.  Tier 6-8.  Tier 9/10.  Not quite that clear cut as 9 almost belongs in the 6-8 and 6s can be more like 4-5, but baby steps....

 

My main question is to contribute to WN8, is it best to comment here, in the main forum thread or elsewhere?

 

Also I can present my hardcore-number crunching skills/statistics abuse skills as needed--I'm a much better scientist/statistician (scientists actualy are pretty bad at stats) than tanker!


  • 0




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users