Jump to content


Real stats repository


  • Please log in to reply
80 replies to this topic

gallabru #41 Posted 04 September 2019 - 07:52 AM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

More a fun fact rather than a stat., but something that was puzzling me concerning the 'bootcamp' for underperforming players : was it the same queue as the newbies' one ?

 

Same sample, I have filtered the players with more than 5k battles and checked their average WR.

> 5k : 102 players, 46.7 % wr

> 7k : 63 players, 46.4% wr

>9k : 56 players, 47% wr

>15k : 30 payers, 48.5% wr

...

Newbie queue is not bootcamp so far.


Edited by gallabru, 04 September 2019 - 10:22 AM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


gallabru #42 Posted 04 September 2019 - 10:23 AM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

updated with 73 battles (949 players). Less noisy but same results

 

Posted Image

 

One spurious point at 63% who is from my toonmate.

 

Here is the average WR of the green and red teams as a function of the battles.

Posted Image


Edited by gallabru, 16 September 2019 - 12:12 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


szkaraj #43 Posted 04 September 2019 - 12:13 PM

    Senior Sergeant

  • Player
  • 43196 battles
  • 798
  • [AFK]
  • Member since:
    05-16-2015
Interesting work. Thx for it. Will check back later again.

The easy day was yesterday.


I_C4NT_W1N_ #44 Posted 04 September 2019 - 12:19 PM

    First Sergeant

  • Player
  • 47972 battles
  • 1,341
  • [1--UK]
  • Member since:
    10-16-2015
Very interesting good work.....

jylpah #45 Posted 04 September 2019 - 05:43 PM

    First Sergeant

  • Player
  • 23068 battles
  • 5,257
  • [AFK]
  • Member since:
    08-05-2014

View Postgallabru, on 11 February 2019 - 12:50 PM, said:

Amongst stats that are nor related to me, here are the WR distributions of my teams and red teams on my alt account. My WR was not included otherwise there would be a peak of 126 at 61.

Both were fitted to a Gaussian. Both are finally centered in 47%. FWHM are 6.4% for my teams and 5.3% for green team, but again, my there are less data for my teams.

If I take all WR of all players (minus me), here is what I get :

The Gaussian is centered in 47% and the width is 5.8%.

 

I once did estimate based on 89k players in BlitzStars' database (with courtesy of SockRobber  :great:  ):

 

WR mean=48.88755000%

WR sigma=5.989095% (Std dev)

 

this estimate is from 2017 if I recall it correctly. Once I get time to update my scripts I will calculate the stats again based on recent data.

 

Anyway, simply awesome stuff!!!

 


Edited by jylpah, 04 September 2019 - 05:48 PM.

Blitzanalysiz.com  In-depth analysis of WoT Blitz game statistics

Jylpah @EU  Jylpah @Asia  Blitz-tools  


Mjr_Leicht #46 Posted 04 September 2019 - 09:40 PM

    Sergeant

  • Player
  • 8095 battles
  • 461
  • [BT_SV]
  • Member since:
    01-21-2019

I expect we'll not see much difference to Jylpah's 2017 analysis but maybe more good players have left as is oft claimed on the forum...

It will be interesting to see some real data....


560273467.png

gallabru #47 Posted 18 September 2019 - 11:42 AM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

Back with some tentative of analysis on how the "improved MM" may or may not work.

I have had to do a lot of assumptions and approximations;

First big thanks to @Jylpah who provided me with the stats he has been collecting for all the players for all their tanks, until update 6.1 included.

 

So here I am with these data, my 73 games on my reroll in 6.2 (4k games, 61%WR) and my questions about improved MM. How coull the MM select the inefficient players ? 

-WR

-DPB

-XP per battle

- hits per battle

 

1/ let's assume that the stats of the players until 6.1 are similar to how they have been playing in 6.2.This is a strong hypothesis.

2/ let's narrow the data to tier 7 and 8 in med tanks. There should still be plenty of data, and some of the relevant parameters should not be too sprayed out (133375).

3/ then let's compare the stats of the players I have seen since 6.2 (222) to those of the player base until then...

Here is the WR distribution of this sample.

Posted Image

 

Here are the stats of the player base until 6.1 (heat map), WR against XP per game. Blue not many players to yellow many players. Let's add the ones I have seen (red crosses).

Posted Image

Same but WR against Damage dealt per tier and per game.

Posted Image

Same but WR against Hits per game.

Posted Image

 

The players I have seen do not seem to depart too much from the total playerbase. Many reasons may explain this :

1/ my sample is small.

2/ the removal of inefficient players did not have time to be in action.

3/ the new MM may not apply to the newbie queue, but the official WG discord says the inefficient players are sent to 'the newcommers queue', so I should see more of them.

4/ I am an inefficient player.

 


Edited by gallabru, 18 September 2019 - 05:11 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


Mjr_Eazy #48 Posted 18 September 2019 - 12:48 PM

    First Sergeant

  • Player
  • 32729 battles
  • 7,597
  • [BT_SV]
  • Member since:
    04-26-2017

Nice one....

 

Your stats are pretty good, not many battles recently but still put you towards the top of the pile by WR & WN8 but your Avg DMG, DR, KDR and Spots are pretty low for the average tier so maybe it's something around DR or KDR vs average for tier which decides efficiency and that would put you in the bootcamp queue?  

 

I don't think it's necessarily newcomers, as WG described the new 6.2 MM as being about have a queue for less efficient players which should mean seal clubbing stat padders like my re-roll don't get in this inefficient players queue but I see more better players which would anecdotally explain my WR plunging off a cliff...

although grinding IS-3 stock to get an IS-8 then grinding that stock to get an IS-7 in the 7-days of the event certificates didn't help as I ignored WR, steamed in, did as much dmg as possible and went to next battle as was flipping between phone and mac to play main account at same time to grind tier ix batchat and E100 :)


547146227.png

 


gallabru #49 Posted 18 September 2019 - 01:00 PM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016
The last message by RibbleStripe of WG Discord do says (I quote) : "players showing extremely low efficiency are sent to newconners queue"

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


Mjr_Eazy #50 Posted 18 September 2019 - 06:25 PM

    First Sergeant

  • Player
  • 32729 battles
  • 7,597
  • [BT_SV]
  • Member since:
    04-26-2017

Interesting, as that isn’t how it was described in release, they said it was for inefficient players, it wasn’t described as having a newcomers queue and inefficient players getting put in that...

but that sounds like they have the sub-5k battles players type queue still and use that, I think the battle count split is a lot lower tho, as my re-roll has seen very high battle count players for a long time and bumped in to some people from here with higher than 5k battles in 6.2...


547146227.png

 


gallabru #51 Posted 18 September 2019 - 07:42 PM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016
I never found their description very clear concerning the queue they would be put in.
The 'test' will carry on in 6.3 according to RibbleStripe. This tells to me that indeed not much happened in 6.2.

I love the irony of this. We were sold the fact that the newbies were separated from the dirty sealclubbers to have time to learn in a safe environment. From now on, to improve their experience, they are sent the very worst experienced players that can be found.

Edited by gallabru, 18 September 2019 - 07:42 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


jylpah #52 Posted 18 September 2019 - 10:58 PM

    First Sergeant

  • Player
  • 23068 battles
  • 5,257
  • [AFK]
  • Member since:
    08-05-2014

View Postgallabru, on 18 September 2019 - 09:42 PM, said:

I never found their description very clear concerning the queue they would be put in.
The 'test' will carry on in 6.3 according to RibbleStripe. This tells to me that indeed not much happened in 6.2.

I love the irony of this. We were sold the fact that the newbies were separated from the dirty sealclubbers to have time to learn in a safe environment. From now on, to improve their experience, they are sent the very worst experienced players that can be found.

 

 

Blitz WWW pages say:

 

https://wotblitz.eu/en/news/updates/update-6-3/

 

New Matchmaker Rule

Testing of the matchmaker rule for players who show extremely low efficiency in recent series of battles has been extended for CIS and EU live servers.

 

 


Blitzanalysiz.com  In-depth analysis of WoT Blitz game statistics

Jylpah @EU  Jylpah @Asia  Blitz-tools  


gallabru #53 Posted 19 November 2019 - 04:05 PM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

Some useless stats of ratings during the last season (September/October).

I have been to Blitzstars and picked the rating of multiple players and their rankings associated with rating. The lowest I have found was 1184 rating and was ranked 74999. I have assumed that 76000 players had a rating to evalutae the fraction of players with at least a given rating.

Here is the plot.

Posted Image

To put some figures, the average/median rating was at 2810. The characteristic spread is 600, meaning that 68% of the player base has a rating comprised between 2210 and 3410. A rating of 4000 makes you better than 94% of the player base. 

From Jylpah's databases, I have identifier 270k active accounts in approximately the same period of time, meaning that 28% of the player base did at least 10 rating battles during the course of the season.


Edited by gallabru, 19 November 2019 - 04:17 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


Mjr_Eazy #54 Posted 19 November 2019 - 05:30 PM

    First Sergeant

  • Player
  • 32729 battles
  • 7,597
  • [BT_SV]
  • Member since:
    04-26-2017
I’m suprised there’s that many people played 10 or more ratings battles, can you see how many play more than 20 as I have a suspicion a lot get a rating and then don’t play...

547146227.png

 


llo1 #55 Posted 19 November 2019 - 07:03 PM

    First Sergeant

  • Player
  • 30886 battles
  • 6,529
  • [1RTR]
  • Member since:
    06-27-2014

I for one don't usually play after rating.....but I now do rating exclusively whilst in a hotel using dodgy wifi!

So when I don't travel, I rarely use Rating.....Same with the Realistic mode, on dodgy hotel wifi I use the Realistic mode when available....



gallabru #56 Posted 19 November 2019 - 09:17 PM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

Not sure if we can know statistically how many battles each one plays in ratings.

I suspect too that many only play ten. And also that many try to reach a value and then stop.


Edited by gallabru, 19 November 2019 - 09:17 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


Mjr_Eazy #57 Posted 20 November 2019 - 06:12 AM

    First Sergeant

  • Player
  • 32729 battles
  • 7,597
  • [BT_SV]
  • Member since:
    04-26-2017

View Postgallabru, on 19 November 2019 - 09:17 PM, said:

Not sure if we can know statistically how many battles each one plays in ratings.

I suspect too that many only play ten. And also that many try to reach a value and then stop.


That’s what I thought but wasn’t certain...

The distribution is interesting still...

My rating usually wanders around between 2.8k and 3.4k got it to 3.6k once - currently I’m yo-yo-ing between silver and gold leagues as around 3k +/- 100 which fits with me being a distinctly average player :)


547146227.png

 


llo1 #58 Posted 20 November 2019 - 05:55 PM

    First Sergeant

  • Player
  • 30886 battles
  • 6,529
  • [1RTR]
  • Member since:
    06-27-2014

View PostMjr_Eazy, on 20 November 2019 - 06:12 AM, said:


That’s what I thought but wasn’t certain...

The distribution is interesting still...

My rating usually wanders around between 2.8k and 3.4k got it to 3.6k once - currently I’m yo-yo-ing between silver and gold leagues as around 3k +/- 100 which fits with me being a distinctly average player :)


Surely by WG "normalisation " a distinctly average player I meet in the Ratings would be rated at 40% judging by the Calibre  of players I see in Rating battles! Even some of the higher "Gold" players simply do not have either enough experience or wr to justify a Gold rating, whereas some of the "Silver" Rated players I have met are over 50% wr and really know how to play......

Showing that Rating has nothing to do with your own experience and skill, and only due to the total lack of decent players on your team!

 

If you do great, and the team does bad, and the game is a loss....you are simply tarrred with the same downgrade brush.....whereas a player that is totally useless, and carried by the team to a win gets a higher rating, and an increase in Rating score..



gallabru #59 Posted 20 November 2019 - 07:50 PM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016
It is true that in some respect rating is a measure of the time you have spent playing rating battles.
Still, games with an average rating at 3400 are more enjoyable than those at 2700. Teams are much more reliable (both reds and greens). Teams at 2700 average rating are completely chaotic.

Edited by gallabru, 20 November 2019 - 07:50 PM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.


gallabru #60 Posted 24 January 2020 - 11:04 AM

    Senior Sergeant

  • Player
  • 18101 battles
  • 675
  • [LAGER]
  • Member since:
    09-01-2016

Update of the season...

Rating distribution (fetched by hand on Blitzstars, so, quite a limited sample) at this date.

Posted Image

This is basically the integral of a gaussian distribution (repartition function ?) centred in 2811 with a characteristic width of 610, meaning that 68% of the payer base hase a rating comprised between 2200 and 3420.

I have also taken the WR of the players I have found.

Posted Image

The orange dot is the rating of 2811, corresponding to a WR of 46.44%, well below the average rating of the player base.

The highest rating (5857) corresponds to that of the leader of the season at this date who has a WR of 62.4%.


Edited by gallabru, 24 January 2020 - 11:07 AM.

My team stats in real time  : http://forum.wotblit...ats-repository/

Carlsberg clan family is recruiting. Contact us.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users