Jump to content


personal rating - what is good one?


  • This topic is locked This topic is locked
49 replies to this topic

RodneyDangerfield #21 Posted Oct 03 2013 - 16:35

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View Postkinger2005, on Oct 03 2013 - 02:04, said:

Meaningless. Since it takes into account battles played.

If battles played is meaningless why do most people get better the more they play, why do teams practice?

Dlur #22 Posted Oct 03 2013 - 16:49

    Captain

  • Players
  • 26480 battles
  • 1,233
  • Member since:
    09-24-2012
All of the current rating systems are more or less rubbish and are only useful if you want to have a quick, easily accessible way to roughly compare a set of players, like when a battle screen is loading through XVM for the purposes of determining which players on your team to support most and which players on the enemy team to be most cautious of.  

Of the current rating systems the noobmeter PR and WN7 are the most reliable metrics, but they still have significant flaws.  Efficiency is all but dead and useless because it can be so easily padded through methods that have little or no impact on ability to win battles.  The wargaming personal rating has significant flaws as well considering that it takes into account the number of battles you play.

If you truly feel the need to compare players you need to look at winrate combined with damage done relative to tier and look at individual performance in specific tanks.  You also need to consider the types of tanks those people are playing (light tanks, heavies, mediums, TDs).  Experience per game is also a decent factor to consider in an analysis.  You can't look at winrate alone because winrate is easy to pad by doing tank companies and seal clubbing.  You can't look at damage alone because some folks pad damage to the point of disregarding the win, plus there are folks that play a lot of light tank that simply enable damage without necessarily doing a lot of it themselves (which is a stat that WoT does not provide in the API).

WN8 should be a step in the correct direction because it attempts to take into account performance in specific tanks, not just overall.  It is still in development, however.  Ultimately the community won't be able to develop a truly solid rating system unless wargaming further opens up their statistical API to include things such as damage assisted (upon spotting/tracking), potential damage, and more.

OuttaGum #23 Posted Oct 03 2013 - 17:01

    Sergeant

  • Players
  • 19795 battles
  • 108
  • Member since:
    08-09-2013
If I could have XVM give me one number, I would pick (Damage Dealt / Damage Received).

Fitz506 #24 Posted Oct 03 2013 - 17:44

    Major

  • Players
  • 49251 battles
  • 2,915
  • [WOPR] WOPR
  • Member since:
    04-23-2013

View PostRodneyDangerfield, on Oct 03 2013 - 16:35, said:

If battles played is meaningless why do most people get better the more they play, why do teams practice?

That's assuming someone is actually playing better with more games. No single number is going to tell you how good a player is. Will it differentiate between the horribles and a unicum? Obviously, but for what we have right now, combo of WR, WN7 and average XP?

GrimJahk #25 Posted Oct 03 2013 - 18:01

    Captain

  • Players
  • 13124 battles
  • 1,081
  • Member since:
    01-14-2013

View PostRodneyDangerfield, on Oct 03 2013 - 16:35, said:

If battles played is meaningless why do most people get better the more they play, why do teams practice?

The reason it's meaningless is that since battles played is a factor, a player with 30k games played and a 47% global WR has a significantly higher Personal Rating than a player with a 54% Global WR (in the same tiers) with only 6k Battles. Yet we both know which player we want on our team.

RodneyDangerfield #26 Posted Oct 03 2013 - 18:12

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View PostGrimJahk, on Oct 03 2013 - 18:01, said:

The reason it's meaningless is that since battles played is a factor, a player with 30k games played and a 47% global WR has a significantly higher Personal Rating than a player with a 54% Global WR (in the same tiers) with only 6k Battles. Yet we both know which player we want on our team.

But if you have a player at 30k games with a 54%wr and a player with 6k games at 54%wr (in the same tier) who would you prefer to have on your team?


GrimJahk #27 Posted Oct 03 2013 - 18:25

    Captain

  • Players
  • 13124 battles
  • 1,081
  • Member since:
    01-14-2013
Your

View PostRodneyDangerfield, on Oct 03 2013 - 18:12, said:

But if you have a player at 30k games with a 54%wr and a player with 6k games at 54%wr (in the same tier) who would you prefer to have on your team?

Well duh... Platoon both.

But, I was answering your question about why people believe Personal Rating is broken regarding the weight placed on number of games. I have no hard data, but the weight placed on games played could be as bad a a player with a 40% WR and 30K games has a higher PR than a 65% WR with 8K. Simply because he has more total wins... 12k vs 5.2K

Almost like you get points added per win. (Congratulations Jimmy! You showed up for school today! You get a star!) Those type of points are awarded to everyone on the winning team regardless of performance.

So a chronic AFK bot spammer that has tanks in 5 games at time will grow his PR faster than a 75% WR player that actually plays. THUS, ironically, if AFKBot#3 just happened to be in with ACE75% and ACE75% carries the match, AFKBot#3 gets the same reward for doing nothing.

_Dreadnaught_ #28 Posted Oct 03 2013 - 18:27

    Major

  • Players
  • 34163 battles
  • 2,415
  • Member since:
    08-10-2011

View Postdrewivy, on Oct 03 2013 - 02:16, said:

for what its worth im over 6000

For a closer barometer of the WG winrate. He ^ has over 6,000 while I have 5,864~. Hope that helps somehow.

_Dreadnaught_ #29 Posted Oct 03 2013 - 18:28

    Major

  • Players
  • 34163 battles
  • 2,415
  • Member since:
    08-10-2011

View PostRodneyDangerfield, on Oct 03 2013 - 18:12, said:

But if you have a player at 30k games with a 54%wr and a player with 6k games at 54%wr (in the same tier) who would you prefer to have on your team?

Agreed

RodneyDangerfield #30 Posted Oct 03 2013 - 18:42

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View PostGrimJahk, on Oct 03 2013 - 18:25, said:

Your Well duh... Platoon both.But, I was answering your question about why people believe Personal Rating is broken regarding the weight placed on number of games. I have no hard data, but the weight placed on games played could be as bad a a player with a 40% WR and 30K games has a higher PR than a 65% WR with 8K. Simply because he has more total wins... 12k vs 5.2KAlmost like you get points added per win. (Congratulations Jimmy! You showed up for school today! You get a star!) Those type of points are awarded to everyone on the winning team regardless of performance. So a chronic AFK bot spammer that has tanks in 5 games at time will grow his PR faster than a 75% WR player that actually plays. THUS, ironically, if AFKBot#3 just happened to be in with ACE75% and ACE75% carries the match, AFKBot#3 gets the same reward for doing nothing.
I was just trying to make a point. Battles played has meaning.  The Russians use bronysite rating and I believe this takes into account battles played too.

Curgin #31 Posted Oct 03 2013 - 18:58

    Sergeant

  • Players
  • 16231 battles
  • 242
  • [-_-] -_-
  • Member since:
    11-23-2011

View PostRodneyDangerfield, on Oct 03 2013 - 16:35, said:

If battles played is meaningless why do most people get better the more they play, why do teams practice?

M#ck#y###, H#ll#e, G#mer##### and many others with tens of thousands of battles and no real skill improvement.  There may be minor correlation between # of battles and skill, but not causation.  I understand that on any non-reroll account, it takes a couple thousand battles to learn how to play better than a good bot.  It does not take the 20k battles WG seems to think it does, or even the 13k for 90% of your score to count.

Spotting damage is the one big thing WG could have brought to the table, as they don't release it in the API.  They didn't, and instead tried to make bad players feel less bad so they would keep spending money.

GrimJahk #32 Posted Oct 03 2013 - 19:15

    Captain

  • Players
  • 13124 battles
  • 1,081
  • Member since:
    01-14-2013

View PostRodneyDangerfield, on Oct 03 2013 - 18:42, said:

I was just trying to make a point. Battles played has meaning.  The Russians use bronysite rating and I believe this takes into account battles played too.
I got your point, but in essence you reinforced mine. The 30Kplayer with the 47%WR and the high PR was summarily dismissed to offer the lower PR 6Kplayer with 54% WR.

Ok, not sure here,

Are you defending/supporting the "every runner gets a medal because they are special" mentality? It would amaze me if the russians would embrace such a system.

GrimJahk #33 Posted Oct 03 2013 - 19:18

    Captain

  • Players
  • 13124 battles
  • 1,081
  • Member since:
    01-14-2013

View PostCurgin, on Oct 03 2013 - 18:58, said:

Spotting damage is the one big thing WG could have brought to the table, as they don't release it in the API.  

^ THIS

Scouts need love too. Especially after sticking AMX 45 up against Tier 10s. you get 5k spotting damage and all you get is extra pay. no stats recognition.

SumiXam #34 Posted Oct 03 2013 - 19:22

    Major

  • Players
  • 7675 battles
  • 4,839
  • Member since:
    09-28-2011

View PostRodneyDangerfield, on Oct 03 2013 - 18:42, said:

I was just trying to make a point. Battles played has meaning.  The Russians use bronysite rating and I believe this takes into account battles played too.

Only to a certain point though. After a few thousand battles, battle count becomes more of diminishing returns factor.

RodneyDangerfield #35 Posted Oct 03 2013 - 19:26

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View PostGrimJahk, on Oct 03 2013 - 19:15, said:

I got your point, but in essence you reinforced mine. The 30Kplayer with the 47%WR and the high PR was summarily dismissed to offer the lower PR 6Kplayer with 54% WR. Ok, not sure here,Are you defending/supporting the "every runner gets a medal because they are special" mentality? It would amaze me if the russians would embrace such a system.

The problem with WGR is it rewards battles played without context.  I believe their rating could work if they reward consistent good long term play, but don't reward consistent long term potatoing.

Curgin #36 Posted Oct 03 2013 - 19:27

    Sergeant

  • Players
  • 16231 battles
  • 242
  • [-_-] -_-
  • Member since:
    11-23-2011

View PostRodneyDangerfield, on Oct 03 2013 - 18:12, said:

But if you have a player at 30k games with a 54%wr and a player with 6k games at 54%wr (in the same tier) who would you prefer to have on your team?

The guy with 6k games.  Every time.  Chances are high that his recent play is well above his overall numbers.  The guy with 30k battles is not likely to have made noticeable recent improvement.

RodneyDangerfield #37 Posted Oct 03 2013 - 19:38

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View PostSumiXam, on Oct 03 2013 - 19:22, said:

Only to a certain point though. After a few thousand battles, battle count becomes more of diminishing returns factor.

Yes and no.  For overall gameplay maybe but for each individual tanks no.  I would say that tazilon knows the capabilities of the 28.01 better than anyone, others might have better stats but he knows the tank inside and out on every map.

RodneyDangerfield #38 Posted Oct 03 2013 - 19:41

    Major

  • Players
  • 72570 battles
  • 6,106
  • Member since:
    04-29-2011

View PostCurgin, on Oct 03 2013 - 19:27, said:

The guy with 6k games. Every time. Chances are high that his recent play is well above his overall numbers. The guy with 30k battles is not likely to have made noticeable recent improvement.
oh like my solo numbers?

Fitz506 #39 Posted Oct 03 2013 - 19:59

    Major

  • Players
  • 49251 battles
  • 2,915
  • [WOPR] WOPR
  • Member since:
    04-23-2013

View PostGrimJahk, on Oct 03 2013 - 19:18, said:

^ THIS

Scouts need love too. Especially after sticking AMX 45 up against Tier 10s. you get 5k spotting damage and all you get is extra pay. no stats recognition.

Spotting damage would be reflected from average XP, no?

SumiXam #40 Posted Oct 03 2013 - 20:09

    Major

  • Players
  • 7675 battles
  • 4,839
  • Member since:
    09-28-2011

View PostRodneyDangerfield, on Oct 03 2013 - 19:38, said:

Yes and no.  For overall gameplay maybe but for each individual tanks no.  I would say that tazilon knows the capabilities of the 28.01 better than anyone, others might have better stats but he knows the tank inside and out on every map.

Absolutely. In fact, it's cases like Taz's (and a host of others in between) that make battle count almost impossible to use as a gauge because the relative meaning of it is highly subjective to many different variables. Different variables will impact individual players in different ways. Battle count is a lot like hit ratio. It doesn't mean a whole lot without other data to provide context to the number.

This is where I tend to somewhat agree with Tazilon's less-than-stellar approval of aggregate stats. There are so many factors that either can't be measured, or measured well, that trying to aggregate everything into a single number will invariably disenfranchise some subset of the data.




1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users