Jump to content


BlitzAnalysiz.com - In-depth analysis of Blitz statistics

BlitzAnalysis.com new site stats statistics

  • Please log in to reply
45 replies to this topic

Jylpah #41 Posted 15 August 2020 - 12:41 AM

    Sergeant

  • Member
  • 19754 battles
  • 184
  • [SKUX]
  • Member since:
    08-06-2015

BIG update: Tank-specific pages added

 

Big update! Tank-specific pages have been added to the site. The pages show detailed analysis of each tier V-X tank. One of the most interesting graphs is the Relative WR for different player skill-levels. WR at the tier is used as a measure for players’ skill. It tells how well the tank is performing for different skill-level players relative to the other players’ other same tier tanks. If the Relative WR is high for players with low WR at the tier, it means the tank is easy to play and less-skilled players do in average better in it than other same tier tanks. These are usually well-armored tanks with high-alpha guns.

 

E 100

 

If the relative WR is negative for players with low WR at the tier, but high for players with high WR at the tier, it means the tank is difficult to play, but can perform well in the hands of a good player. These are often tanks that require high-level of tactical skills and gameplay. It might be frustrating for less-skilled players to play these tanks.

 

T-62A

Stats levels vs. WR in a tank

Tank pages also have graphs showing input-stat levels (average damage, average kills, spot rate, etc.) for different WR levels. The bars show 20-80% of the players within the WR bucket (20% have lower stats and 20% have higher stats with the same WR).

E 100: Average damage vs . WR

X/Y Stats vs. WR in a tank

Finally, there are contour-plots of two input-stats vs. WR. These graphs show what combination of stats are require in average to reach certain WR with a tank.

E 100 - Average Damage + Kills vs. WR

Please note that the plotting library tries to make the contour lines convex and therefore certain graphs have odd behaviors on the edges.


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


Jylpah/blitz-tools      Python tools to upload and analyze replays
Jylpah[SKUX] @Asia       Jylpah[AFK] @EU       

 


erwin10001 #42 Posted 15 August 2020 - 05:27 AM

    Captain

  • Member
  • 49240 battles
  • 1,267
  • [NETS]
  • Member since:
    11-12-2016

this is really awesome in terms of understanding each tank. 

 

i also like the chart that shows how many players have less than 30% winrate and 1k+ games.  these are clearly all trolls that should be banned immediately. (1000 games afk, you would still get more than 30% winrate - to get under 30% winrate after 1k games, you have to be deliberately trying to make your team lose). but of course, WG is not going to do anything about it.


Edited by erwin10001, 15 August 2020 - 05:28 AM.


Jylpah #43 Posted 15 August 2020 - 10:20 AM

    Sergeant

  • Member
  • 19754 battles
  • 184
  • [SKUX]
  • Member since:
    08-06-2015

View Posterwin10001, on 15 August 2020 - 07:27 AM, said:

this is really awesome in terms of understanding each tank. 

 

i also like the chart that shows how many players have less than 30% winrate and 1k+ games.  these are clearly all trolls that should be banned immediately. (1000 games afk, you would still get more than 30% winrate - to get under 30% winrate after 1k games, you have to be deliberately trying to make your team lose). but of course, WG is not going to do anything about it.

 

Either trolls or pre-schoolers :-)

 

Just a note that the site does not have _all_ Blitz players in its database and it is more likely to for an active player at the higher tiers to end up in the database. This is the same thing with BlitzStars.com etc. There is no WG API call to "list all the active players", but one has to "find" the account_ids. Blitzanalysiz collects those from replays. Blitzstars.com gets to know an account once someone searches for the account from the site.

 

Blitz seems to have quite an churn issue. There are huge number of players with little battles and I am sure BlitzAnalysiz knows only small share of those. That is why the average WR is 50%+ which should be "impossible". Those Tiers/tier histograms of "all players" give you perspective of how many of those inactive players there are. In  reality there are much more of those.

 


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


Jylpah/blitz-tools      Python tools to upload and analyze replays
Jylpah[SKUX] @Asia       Jylpah[AFK] @EU       

 


Jylpah #44 Posted 28 August 2020 - 11:11 AM

    Sergeant

  • Member
  • 19754 battles
  • 184
  • [SKUX]
  • Member since:
    08-06-2015

Stats for update 7.1 are (finally) out.Tank-specific are available since 7.0 and now for all the tanks. After adding the tank-pages the site build time has grown 10x and I have been looking for parallel processing to speed up the work. But I have ran into weird issues likely related to shared memory limits causing the parallel R build script to fail. :arta:

 

The graph below shows the strongest tanks in game tier-for-tier

 


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


Jylpah/blitz-tools      Python tools to upload and analyze replays
Jylpah[SKUX] @Asia       Jylpah[AFK] @EU       

 


Newb179 #45 Posted 30 August 2020 - 02:12 AM

    Staff Sergeant

  • Member
  • 9544 battles
  • 304
  • [CPU]
  • Member since:
    07-18-2018
Great work

Jylpah #46 Posted 11 October 2020 - 01:16 PM

    Sergeant

  • Member
  • 19754 battles
  • 184
  • [SKUX]
  • Member since:
    08-06-2015

2020-10-11 Update 7.2 stats are out

  • Record 205M battles by 456k players was recorded in the database. Please note, these are note all the WoT Blitz players, just those tracked by BlitzAnalysiz().
  • There seem to be more low-tier stat-padders captured in the database since Top 5 highest WR tanks are Tier V or below.

Top 10 tanks by Average WR

  • T28 Concept, Object 252U and Chimera continue to be the best performing tanks in the game, with T54E2 taking the 3rd place in front of Chimera. Top four tanks are all premium tanks. It must be a pure accident by WG balancing department. If it were a result of conscious decision, someone could say the game is becoming pay-to-win…

Best performing tanks, update 7.2

Best performing tier X tanks

  • And I finally managed to fix the annoying parallel processing bug that has taken most of my BlitzAnalysiz development time. The issue was that multi-threaded OpenBLAS-libraries used by loess() not being thread-safe in multi-core applications. library('RhcpBLASctl';) ; blas_set_num_threads(1) took care of it. It took probably 50h to debug it. Debugging parallel code is PIA. Finally solved it! By definition, bugs are found in the last place you are looking for those.

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


Jylpah/blitz-tools      Python tools to upload and analyze replays
Jylpah[SKUX] @Asia       Jylpah[AFK] @EU       

 





1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users