Jump to content


NoobMeter.com - WoT performance rating (player comparison tool)

noobmeter performance rating efficiency noobmeter.com performance rating efficiency rating

  • Please log in to reply
1466 replies to this topic

FailAccount #621 Posted 25 March 2013 - 10:15 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012
The thread is up :)

NoobMeter #622 Posted 25 March 2013 - 10:18 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostProfessionalProgamerer, on 25 March 2013 - 10:07 PM, said:

That's cool but how Is IDEALs win rate lower (for example)?

i ask because the members have asked before and we are all registered.

I think it is because for example if I have a 30 day snapshot for you available at www.noobmeter.com, but only have run local datamine runs once every 2 months then it would consider your 30 day interval as the one at www.noobmeter.com but a 2-month interval in this report. So if you played in March better than in February, you can get a difference.

It is not a very good explanation, but the best I can think of at the moment.

But it is not nice. You can hold on with posting it elsewhere, I will download the many-GB www.noobmeter.com DB backup to my home system, restore it there, then run a full datamine again, then create another report and it should be more precise hopefully. But that will take a bit of time :).

If that won't solve it, then I have to check further.

FailAccount #623 Posted 25 March 2013 - 10:20 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012
That's very kind of u. I have already posted the blog but the sanity check can still be done in the background I guess. Not to worry.

NoobMeter #624 Posted 25 March 2013 - 10:22 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostProfessionalProgamerer, on 25 March 2013 - 10:20 PM, said:

That's very kind of u. I have already posted the blog but the sanity check can still be done in the background I guess. Not to worry.

Yes, I posted a disclaimer into the blog post. Thanks for posting the thread and good job on catching this issue so fast. :)

Edited by NoobMeter, 25 March 2013 - 10:22 PM.


FailAccount #625 Posted 25 March 2013 - 10:29 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012
You are most welcome. Everyone appreciates the work you have been doing.

NoobMeter #626 Posted 25 March 2013 - 11:47 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostProfessionalProgamerer, on 25 March 2013 - 10:29 PM, said:

You are most welcome. Everyone appreciates the work you have been doing.
Good to hear that :)

I looked at the IDEAL discrepancies. Basically, the issue is indeed different reporting periods.

For example, for ShiftyShaft a period in January with 59.3% was taken for my local DB (not sure why, maybe his later snapshots had timed out at WG). However his more precise data at www.noobmeter.com has him at 72.40%. Similar for Private_Miros, eliocoric, MAMILYTank, etc.

For other clans this discrepancy maybe wouldn't make a big difference as for some WR/etc would improve, for others worsen, making it a wash.

But somehow IDEA has more people that are doing better in the "more precise" dataset that www.noobmeter.com has.

Also, 1 player left and another joined at end of March, and the databases have different information in that regard.

tl;dr bottom line - the report on the blog isn't as accurate as it could be and doesn't necessarily represent exactly "February 25 - March 24" timeframe for each and every player.

But it should be in the right ballpark and is hopefully still useful.

Edited by NoobMeter, 25 March 2013 - 11:48 PM.


FailAccount #627 Posted 25 March 2013 - 11:54 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012
Thanks for the quick response :)

To make your life easier and more importantly encourage clans to register players, why not simply use the live db? There will be very little discrepancy and if someone feels they could get better results they would need to register the player etc... Which is sort of a win win for u :)

NoobMeter #628 Posted 26 March 2013 - 08:44 AM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostProfessionalProgamerer, on 25 March 2013 - 11:54 PM, said:

Thanks for the quick response :)

To make your life easier and more importantly encourage clans to register players, why not simply use the live db? There will be very little discrepancy and if someone feels they could get better results they would need to register the player etc... Which is sort of a win win for u :)

Yes, I will do that. It does actually make a lot of sense :)

I can maybe even implement it as a feature of the site itself, and not a report that takes my time to run.

Edited by NoobMeter, 26 March 2013 - 08:54 AM.


FailAccount #629 Posted 26 March 2013 - 09:01 AM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012

View PostNoobMeter, on 26 March 2013 - 08:44 AM, said:

Yes, I will do that. It does actually make a lot of sense :)

I can maybe even implement it as a feature of the site itself, and not a report that takes my time to run.
:great:  All the commanders can now obsess about clan stats even more now  :trollface:

NextToYou #630 Posted 26 March 2013 - 01:34 PM

    Staff Sergeant

  • Player
  • 40097 battles
  • 349
  • [EFE] EFE
  • Member since:
    02-07-2011
Feedback (may sound harsh but its just feedback):

The FV215 (183) brit tier10 td formula is totally broken in PR-rating. If I play it exclusively my performance ratings for those day go down to the toilet even thou I really don´t play it worse than other tier10 td´s. I don´t know what kind of average damages one should do to have similar ratings with that compared to other TD´s, but currently you have set the standards for it way too high. In reality, in random battle environment, it will perform worse than best tier10 td´s (mainly foch155 and obj268). I have now driven with it now over 100 battles, so sample size is decentish and in full gold. With other TD´s I only carry a few gold rounds and still my avg dmg is lower in FV215 (183), because it just is worse (atleast worse than those best tier10 tds). But still in PR-rating, standards for it have been set much higher than for other tier10 tds.

In general performance rating favours middle to high tier heavies and mediums and especially derp tanks like M4 Sherman way too high. I cannot use PR-data from it for player evaluation purposes if player has a lot of games played in those middle tier derps. And as been said before, it looks like to diminish tier10 td´s in general (standards are set relativily too high), but especially the new brit tier10 td. Tier10´s are full of arty parties so relative potential dmg is way lower in those games than in lets say tiers 5-9. Also average level of players is higher in tier10 than in lower tiers.

In conclusion: IMO the formulas how to calculate PR for many tanks still need a lot of tweaking. For the whole tank classes and tiers and several individual tanks.

Still the webpage is good, since you can check a lot of data from one page about a player, so gj and ty for that, but for general evaluation of a player WN7 seems to work much better currently (it has its own flaws too, but different kind of and not as big).

Edited by NextToYou, 26 March 2013 - 02:35 PM.


Lille_ #631 Posted 26 March 2013 - 02:11 PM

    Warrant Officer

  • Player
  • 41937 battles
  • 636
  • Member since:
    09-08-2011

View PostSweetheart, on 13 February 2013 - 08:47 AM, said:

I have to say I am beginning to lose faith in noobmeter. Ridiculously high requirements for (some) TDs and ridiculously low for (some) heavies.

http://www.noobmeter...20130213_031750
2.44 kills per game, 3756 DPB, 92,6% WR = 2053 PR
Played with the Foch 155.


http://www.noobmeter...20130213_030124
2.37 kills per game, 3921 DPB, 92,6% WR = 2854 PR
Played mostly with the T57 Heavy.


So, a difference of 165 DPB between a low DPM TD and a high DPM heavy = over 800 PR.

View PostNextToYou, on 26 March 2013 - 01:34 PM, said:

- too long! -

You see there are other examples. IMO there are lots of mistakes in PR thats why I don't like this rating.

Example: How can a player with 50.8 % WR, 0.83 kills/game, average damage of 500 and average tier 4,8 get 1.5k PR? WN and EFF are arround 1k. That's just wrong and does not fit to players stats, cause it's definetely not "good".

I love noobmeter.com, best side for tracking stats. But PR is not good.

NoobMeter #632 Posted 26 March 2013 - 04:27 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostNextToYou, on 26 March 2013 - 01:34 PM, said:

Feedback (may sound harsh but its just feedback):
You have feedback with examples, I'm fine with such.

Quote

The FV215 (183) brit tier10 td formula is totally broken in PR-rating. If I play it exclusively my performance ratings for those day go down to the toilet even thou I really don´t play it worse than other tier10 td´s. I don´t know what kind of average damages one should do to have similar ratings with that compared to other TD´s, but currently you have set the standards for it way too high.
That is very very very strange because it's expected damage is EXACTLY the same as all the other t10 TDs. I just didn't know enough about the tank to set it otherwise.

There must be something else at play there.

Quote

In reality, in random battle environment, it will perform worse than best tier10 td´s (mainly foch155 and obj268). I have now driven with it now over 100 battles, so sample size is decentish and in full gold. With other TD´s I only carry a few gold rounds and still my avg dmg is lower in FV215 (183), because it just is worse (atleast worse than those best tier10 tds). But still in PR-rating, standards for it have been set much higher than for other tier10 tds.

As mentioned, standards have been set the same for all t10 TDs (though they are higher than for t10 HTs or meds).

If I get consistent feedback from great players like you that some TDs should have lower expected damage than others, then I'm fine with adjusting it.

Assume expected damage for JgPzE-100 is 2500 . What would you set the expected damage for all the others?

Quote

In general performance rating favours middle to high tier heavies and mediums and especially derp tanks like M4 Sherman way too high. I cannot use PR-data from it for player evaluation purposes if player has a lot of games played in those middle tier derps.
The gold ammo derp spam is a problem.

I could solve it by adding M4 Sherman to OP tanks list and then for people with a lot of battles in it with significantly higher winrate than account winrate it will be judged more harshly.

That will however cause a lot of objections from many people playing it who feel it is not OP.

Again I would welcome feedback from other great players on this.

Quote

And as been said before, it looks like to diminish tier10 td´s in general (standards are set relativily too high), but especially the new brit tier10 td. Tier10´s are full of arty parties so relative potential dmg is way lower in those games than in lets say tiers 5-9. Also average level of players is higher in tier10 than in lower tiers.

In conclusion: IMO the formulas how to calculate PR for many tanks still need a lot of tweaking. For the whole tank classes and tiers and several individual tanks.

I'm ready to accept input on this and tweak it. I'm just not sure how to do it best without it becoming an unmanageable chaos of people suggesting different things.

Quote

Still the webpage is good, since you can check a lot of data from one page about a player, so gj and ty for that, but for general evaluation of a player WN7 seems to work much better currently (it has its own flaws too, but different kind of and not as big).

I had the opposite feedback lately for other examples.

NoobMeter #633 Posted 26 March 2013 - 04:29 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostSchnudi, on 26 March 2013 - 02:11 PM, said:

Example: How can a player with 50.8 % WR, 0.83 kills/game, average damage of 500 and average tier 4,8 get 1.5k PR? WN and EFF are arround 1k. That's just wrong and does not fit to players stats, cause it's definetely not "good".

Which is that player in particular?

But the point is not to compare it 1:1 to WN7 and ER, they have different scales.

If you find a better player who has lower PR or vice versa - then it is a problem.

Edited by NoobMeter, 26 March 2013 - 04:30 PM.


NoobMeter #634 Posted 26 March 2013 - 04:32 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostSweetheart, on 13 February 2013 - 08:47 AM, said:

I have to say I am beginning to lose faith in noobmeter. Ridiculously high requirements for (some) TDs and ridiculously low for (some) heavies.

http://www.noobmeter...20130213_031750
2.44 kills per game, 3756 DPB, 92,6% WR = 2053 PR
Played with the Foch 155.


http://www.noobmeter...20130213_030124
2.37 kills per game, 3921 DPB, 92,6% WR = 2854 PR
Played mostly with the T57 Heavy.


So, a difference of 165 DPB between a low DPM TD and a high DPM heavy = over 800 PR.

I agree with Sweetheart. T57 Heavy probably needs its expectations boosted.

VeryRisky #635 Posted 26 March 2013 - 05:08 PM

    General

  • Player
  • 17126 battles
  • 8,529
  • [WJDE] WJDE
  • Member since:
    10-11-2012

View PostNoobMeter, on 26 March 2013 - 04:27 PM, said:

The gold ammo derp spam is a problem.

I could solve it by adding M4 Sherman to OP tanks list and then for people with a lot of battles in it with significantly higher winrate than account winrate it will be judged more harshly.

That will however cause a lot of objections from many people playing it who feel it is not OP.

Again I would welcome feedback from other great players on this.


Well other than waiting for WG to fix the acknowledged 105 HEAT issue (Just drop the damage - other HEAT shells have much lower damage than HE, this one is only 25% down) the only think I can thing of is having the formula non-liner so you can't get that above a certian level the score/damage decreases (or goes with the sqrt of the damage rather than liner).

NoobMeter #636 Posted 26 March 2013 - 11:54 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012
Added feature of showing clan tops.

Thanks to ProfessionalProgamerer for suggesting it.

FailAccount #637 Posted 27 March 2013 - 02:18 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012
Thank you Mr Noob Meter, the update is fantastic.

Following the release a number of clans are disputing the size of the clans that should be shown.

Arguably a clan isn't a clan if they cant field a full team of 15 in CW or TCs. Some argue that you need 30 players in a team.

Would it be possible to filter out the "micro" clans?


In addition . For the recent clan table can you replace the overall battles column with the number of recent battles.

Many thanks again for your devotion to such a great and valuable project.

NoobMeter #638 Posted 27 March 2013 - 03:37 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012

View PostProfessionalProgamerer, on 27 March 2013 - 02:18 PM, said:

Thank you Mr Noob Meter, the update is fantastic.
Following the release a number of clans are disputing the size of the clans that should be shown.
Arguably a clan isn't a clan if they cant field a full team of 15 in CW or TCs. Some argue that you need 30 players in a team.
Would it be possible to filter out the "micro" clans?

I added filtering by >= 15 players for now. Depending on further feedback, I will consider increasing it to 30.

It is not deployed yet, will deploy when more changes add up (I don't want to disturb the hundreds of users on the system currently for a small change).

Quote

In addition . For the recent clan table can you replace the overall battles column with the number of recent battles.
Many thanks again for your devotion to such a great and valuable project.

I think the last column is already "battles (recent)". Or am I missing something?

Edited by NoobMeter, 27 March 2013 - 03:41 PM.


NoobMeter #639 Posted 27 March 2013 - 10:39 PM

    Warrant Officer

  • Player
  • 1 battle
  • 534
  • Member since:
    11-14-2012
New tank WR/OP report out:

http://blog.noobmete...tes-and-op.html

P.S. Feel free to repost in other forums / threads as you see fit. I no longer have rights to start new threads in most forums due to low amount of battles.

Edited by NoobMeter, 27 March 2013 - 10:42 PM.


FailAccount #640 Posted 27 March 2013 - 10:58 PM

    Captain

  • Player
  • 1960 battles
  • 2,425
  • Member since:
    04-30-2012

View PostNoobMeter, on 27 March 2013 - 03:37 PM, said:

I added filtering by >= 15 players for now. Depending on further feedback, I will consider increasing it to 30.

It is not deployed yet, will deploy when more changes add up (I don't want to disturb the hundreds of users on the system currently for a small change).



I think the last column is already "battles (recent)". Or am I missing something?
Great :) ty.


I see recent battles now. Sure it wasn't there at some point. Either way, I and my wot friends are very grateful. Ty.





Also tagged with noobmeter, performance, rating, efficiency, noobmeter.com, performance rating, efficiency rating

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users