Jump to content


Blitz statistics (Since update v5.4)

Statistics

  • Please log in to reply
48 replies to this topic

llo1 #21 Posted 04 February 2020 - 04:27 PM

    First Sergeant

  • Player
  • 30162 battles
  • 6,304
  • [1RTR]
  • Member since:
    06-27-2014

Hats off to you jylpah....my stats knowledge disappeared into the ether years ago.

But as you say, and taking me as a start, I DO NOT upload replays to the system, unless there is something I want people to view....either good aces, or totally defunct system (in my eyes) ....so if others are doing that would it therefore skew the results?

Or does your data acquisition script simply pull everything down fromWG that they allow you to gather?



jylpah #22 Posted 04 February 2020 - 07:28 PM

    First Sergeant

  • Player
  • 22275 battles
  • 5,124
  • [AFK]
  • Member since:
    08-05-2014

View Postllo1, on 04 February 2020 - 06:27 PM, said:

Hats off to you jylpah....my stats knowledge disappeared into the ether years ago.

But as you say, and taking me as a start, I DO NOT upload replays to the system, unless there is something I want people to view....either good aces, or totally defunct system (in my eyes) ....so if others are doing that would it therefore skew the results?

Or does your data acquisition script simply pull everything down fromWG that they allow you to gather?

 

Thanks llo1,

 

I use the replays only to collect account_ids, nothing more. After that I run the following request towards the WG API.

 

 

You can get your own API key for free by "registering" your app at https://developers.wargaming.net/applications/

The WG_ACCOUNT_ID is a integer account_id

 

This request will return a JSON file like (but for all the player's tanks, I have removed all but one tank record

 

Block Quote

 {
    "status": "ok",
    "meta": {
        "count": 1
    },
    "data": {
        "521458531": [
            {
                "all": {
                    "spotted": 600,
                    "hits": 2684,
                    "frags": 413,
                    "max_xp": 1879,
                    "wins": 231,
                    "losses": 141,
                    "capture_points": 0,
                    "battles": 373,
                    "damage_dealt": 530531,
                    "damage_received": 386121,
                    "max_frags": 5,
                    "shots": 3189,
                    "frags8p": 338,
                    "xp": 314629,
                    "win_and_survived": 151,
                    "survived_battles": 153,
                    "dropped_capture_points": 175
                },
                "last_battle_time": 1579947215,
                "account_id": 521458531,
                "max_xp": 0,
                "in_garage_updated": 0,
                "max_frags": 0,
                "frags": null,
                "mark_of_mastery": 3,
                "battle_life_time": 62195,
                "in_garage": null,
                "tank_id": 62737
            }
        ]
    }
}

 

Please note that WG shares onlyt the current stats, not snapshots from the history. That's why I have to fetch the stats during the update since I am mostly interested in stats over one update and the player base as the whole - not individual player stats. The API is well documented. Just visit https://developers.wargaming.net/

 


Edited by jylpah, 04 February 2020 - 07:28 PM.

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

Jylpah @EU  Jylpah @Asia  Blitz-tools  


Jukkis74 #23 Posted 04 February 2020 - 07:36 PM

    Senior Sergeant

  • Player
  • 34919 battles
  • 823
  • [AFK]
  • Member since:
    06-29-2014

Smallish issue but I would be interested in it: what is the split of lower/higher-tier tanks in different battle tiers?

 

E.g. if top tier is 10, how often there is 0,1,2,...,6 t9 tanks. If top tier is t9, how often there is 0,1,...,6 t8 tanks.

 

I am just wondering how MM handles the bottom/toptier mixing for matches.



jylpah #24 Posted 04 February 2020 - 08:58 PM

    First Sergeant

  • Player
  • 22275 battles
  • 5,124
  • [AFK]
  • Member since:
    08-05-2014

View PostJukkis74, on 04 February 2020 - 09:36 PM, said:

Smallish issue but I would be interested in it: what is the split of lower/higher-tier tanks in different battle tiers?

 

E.g. if top tier is 10, how often there is 0,1,2,...,6 t9 tanks. If top tier is t9, how often there is 0,1,...,6 t8 tanks.

 

I am just wondering how MM handles the bottom/toptier mixing for matches.

 

This would require parsing through the replay files. The database has 65k replays, but the issue is that the replay selection is highly skewed and could cause bias. If one start filtering out Mastery badges there is one introduces another unknown bias.

 

I think the best way to figure out bottom/top tier is to play the same tier tanks a large number of battles, save the replays and run the replay analyzer script. At least there is not bias then when one selects all the battles.


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

Jylpah @EU  Jylpah @Asia  Blitz-tools  


Mjr_Eazy #25 Posted 04 February 2020 - 09:47 PM

    First Sergeant

  • Player
  • 31860 battles
  • 7,275
  • [BT_SV]
  • Member since:
    04-26-2017

View PostJukkis74, on 04 February 2020 - 07:36 PM, said:

Smallish issue but I would be interested in it: what is the split of lower/higher-tier tanks in different battle tiers?

 

E.g. if top tier is 10, how often there is 0,1,2,...,6 t9 tanks. If top tier is t9, how often there is 0,1,...,6 t8 tanks.

 

I am just wondering how MM handles the bottom/toptier mixing for matches.

 

View Postjylpah, on 04 February 2020 - 08:58 PM, said:

 

This would require parsing through the replay files. The database has 65k replays, but the issue is that the replay selection is highly skewed and could cause bias. If one start filtering out Mastery badges there is one introduces another unknown bias.

 

I think the best way to figure out bottom/top tier is to play the same tier tanks a large number of battles, save the replays and run the replay analyzer script. At least there is not bias then when one selects all the battles.


That’s along the lines of what I did for the P1 vs P2 thread - played around 300 battles in Smasher on 2 accounts but I’ve excluded tier x from data as would skew my top/bottom tier frequency stuff...

That’s the most I’ll have in one tank in one release, so could try with them if you like?

I’ll have a look and see what I’ve got in tier ix/x tanks but it’s gonna span multiple releases...


547146227.png

 


Jukkis74 #26 Posted 05 February 2020 - 05:39 AM

    Senior Sergeant

  • Player
  • 34919 battles
  • 823
  • [AFK]
  • Member since:
    06-29-2014
I would be happy to see the tier statistics for the currently available 65k battle population. More fun to speculate around MM when you have some facts available :-)

jylpah #27 Posted 05 February 2020 - 11:22 AM

    First Sergeant

  • Player
  • 22275 battles
  • 5,124
  • [AFK]
  • Member since:
    08-05-2014

View PostJukkis74, on 05 February 2020 - 07:39 AM, said:

I would be happy to see the tier statistics for the currently available 65k battle population. More fun to speculate around MM when you have some facts available :-)


I gave a bit thought on this and it seems the analyze_wotb_replays.py could be relatively easily modified to read the replays from the database instead of the file system. However, the current app relies on in-memory cache for stats and this might become an issue. Also, the current database has 500k users and 91M stats records. Running full analysis might several hours ( a day ?) to complete. 
 

But I will definitely work on this just to provide another set of skewed analysis for the endless “debate” :harp:
 

Btw. It’s possible to filter replays based on senders’ medals (mastery, 1st class, etc.). Running these would provide additional insights to the question. 


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

Jylpah @EU  Jylpah @Asia  Blitz-tools  


Corpsetaker #28 Posted 05 February 2020 - 12:31 PM

    Sergeant

  • Player
  • 19211 battles
  • 498
  • [PT40]
  • Member since:
    09-07-2016
I think Blitz stars is flawed. I won three battles on my alt account and my WR actually went down.

Buckleyteer2 #29 Posted 05 February 2020 - 12:58 PM

    First Sergeant

  • Player
  • 46452 battles
  • 1,159
  • [-SOF-]
  • Member since:
    10-16-2015
WTF !!!!

Corpsetaker #30 Posted 05 February 2020 - 01:06 PM

    Sergeant

  • Player
  • 19211 battles
  • 498
  • [PT40]
  • Member since:
    09-07-2016

View PostBuckleyteer2, on 05 February 2020 - 12:58 PM, said:

WTF !!!!

Yeah I said the same thing.



Replicant_2 #31 Posted 05 February 2020 - 01:38 PM

    First Sergeant

  • Player
  • 9605 battles
  • 1,027
  • [HRG]
  • Member since:
    10-08-2018

View PostCorpsetaker, on 05 February 2020 - 01:06 PM, said:

Yeah I said the same thing.


Could it be that you were looking at 30d WR? 
As it’s a sliding window type of a snapshot it’s affected by both the newly added records and the ones removed at the tail.

anyway, just a random guess as there’s not enough data :)


All those moments will be lost in time, like tears in rain.

Time to die.


Buckleyteer2 #32 Posted 05 February 2020 - 02:13 PM

    First Sergeant

  • Player
  • 46452 battles
  • 1,159
  • [-SOF-]
  • Member since:
    10-16-2015
Aaaaahhhhhhh

BaronVonPewpew #33 Posted 05 February 2020 - 02:23 PM

    First Sergeant

  • Player
  • 22695 battles
  • 1,293
  • [FTC-1]
  • Member since:
    08-16-2016
So let me get this right, the average win rate at 3.8 was 48% and now it’s improved to 50% average. So does this mean the newer players have improved or what we said at update 5.5 about Wargaming trying to get the bulk of players to 50% as the average has come true through detriment to veteran players who since 5.5 have had a rough ride. I can only go on feedback from our large number of members but what the seasoned players all say unanimously is that they all have been a steady decline in Wr over the last 14 months and over the last 3 months it has accelerated very considerably.

​Very proud leader of the Fun-Tan-Clan 


BaronVonPewpew #34 Posted 05 February 2020 - 02:27 PM

    First Sergeant

  • Player
  • 22695 battles
  • 1,293
  • [FTC-1]
  • Member since:
    08-16-2016

View PostBuckleyteer2, on 05 February 2020 - 12:58 PM, said:

WTF !!!!


This does happen in game too, we only discussed this a few weeks back on discord.


​Very proud leader of the Fun-Tan-Clan 


Buckleyteer2 #35 Posted 05 February 2020 - 02:41 PM

    First Sergeant

  • Player
  • 46452 battles
  • 1,159
  • [-SOF-]
  • Member since:
    10-16-2015
I will second that mine has slumped into disrepair !!!

llo1 #36 Posted 05 February 2020 - 03:38 PM

    First Sergeant

  • Player
  • 30162 battles
  • 6,304
  • [1RTR]
  • Member since:
    06-27-2014

View PostVonPottypoop, on 05 February 2020 - 02:23 PM, said:

So let me get this right, the average win rate at 3.8 was 48% and now it’s improved to 50% average. So does this mean the newer players have improved or what we said at update 5.5 about Wargaming trying to get the bulk of players to 50% as the average has come true through detriment to veteran players who since 5.5 have had a rough ride. I can only go on feedback from our large number of members but what the seasoned players all say unanimously is that they all have been a steady decline in Wr over the last 14 months and over the last 3 months it has accelerated very considerably.


Of course this is true....if you “encourage” all the longer term, better wr players to leave then of course the wr of the average player will increase! And of course if you believe that the aim is to get 50% wr for everyone, removing the better wr from the top end (losing players) and boosting the noobs quickly away from total sealclubbed starters to where they are classed as average (by buying tanks they do not know how to play) and ensuring they cannot be sealclubbed, then the average WILL be nearer 50%.......

In WG’s eyes...job well done!

 



minitelrose #37 Posted 06 February 2020 - 08:22 AM

    First Sergeant

  • Player
  • 78153 battles
  • 8,110
  • [AFK]
  • Member since:
    07-06-2014

View Postjylpah, on 03 February 2020 - 09:22 AM, said:

 

btw. I would appreciate ideas for analysis topics. I will be adding stuff. These are on my task list:

  1. Tank specific performance graphs (e.g. average damage + kills vs WR)
  2. Best/worst tank per tier


1-
i would only do it to keep the official numbers published by WG  in check. They have no reason to cheat, we have no reason to believe them, but there is no reason to have the exact same graphs from 2 different sources every month. That way you could concentrate both your fire and your readers attention on other things. Just a minor remark. On the other hand you might be able to do it better or different from the WG published graphs, in which case I’m curious to see them.

 


2-

i know of like that better, but how to determine best tank ? And best tank for who ?

that is also an issue with what is the purpose of a tank ? Increase dmg numbers ? Increase victory ratios ? Have fun ? Create piles of credits in a short time ? I have little clue how to make a best tank except for better wins (cause otherwise fv183 would be a good tank, and it is not).

Best for who? means that certainly there are tanks that only benefit the lesser skilled players, and tanks that require enormous amount of skills to perform, but that outperform anything when driven well.


as a result I would just make top winrates for each slice of player skills. And maybe per tier?

 

any other thoughts on this, anyone ?


Read my guides there

"I want the game to be just how it was launched, no MM limitations for platoons, some unbalanced match once in a while, and friendly fire should come back as well. The only thing they should remove is statistics." -phony1907, 08 July 2016, 05:45pm - 

I want just that, and replay files.

 


minitelrose #38 Posted 06 February 2020 - 08:30 AM

    First Sergeant

  • Player
  • 78153 battles
  • 8,110
  • [AFK]
  • Member since:
    07-06-2014

View PostChairman_merpug, on 03 February 2020 - 10:19 AM, said:

On point 2: more replays needed, then. Bulk uploading them would be nice. Is there something that I could use on my IPad to just haul them out of whatever folder the thing hides them in and sling them at wotinspector.com? Even then, the games I rage quit because the other "players" are having trouble finding their backsides with both hands and a map would be excluded.


you can. There is no limit to how many per email dump or per discord dump or per direct dump in the website, 

however servers struggle with single dumps passed a certain size.

 

 

inthink it’s not geared to receive every replay. A better solution would be to have a device based replay analysis tool.
it could provide a personal stat monitor system for exemple. 
its  something that andrew has discussed for a long time, but he lacks time / resource / crowd interest on the subject.


Read my guides there

"I want the game to be just how it was launched, no MM limitations for platoons, some unbalanced match once in a while, and friendly fire should come back as well. The only thing they should remove is statistics." -phony1907, 08 July 2016, 05:45pm - 

I want just that, and replay files.

 


minitelrose #39 Posted 06 February 2020 - 08:33 AM

    First Sergeant

  • Player
  • 78153 battles
  • 8,110
  • [AFK]
  • Member since:
    07-06-2014

 

Edited :

Solid evidence of forum bug here.


Edited by minitelrose, 06 February 2020 - 08:33 AM.

Read my guides there

"I want the game to be just how it was launched, no MM limitations for platoons, some unbalanced match once in a while, and friendly fire should come back as well. The only thing they should remove is statistics." -phony1907, 08 July 2016, 05:45pm - 

I want just that, and replay files.

 


minitelrose #40 Posted 06 February 2020 - 08:41 AM

    First Sergeant

  • Player
  • 78153 battles
  • 8,110
  • [AFK]
  • Member since:
    07-06-2014

View PostMjr_Eazy, on 04 February 2020 - 12:11 AM, said:

 

Interesting stuff Jylpah, as you say it would make sense there’s a bias to wins in the replay data as people more likely to upload wins/aces/good battles...

 

 

 

Indeed that might be the main bias here, at least a strong one :


most replays submitted are victories, and about 50% might be aces.

 

You know that’s not nearly representative of games. That would mean in every other game you play someone in either team aced his thing. Also that would mean most games both teams would be victorious and anyone would rarely be defeated.

 

also, I suspect only the most well informed players submit replays. Of course there has to be noobs in the lot of uploaders, but most certainly noobs that have had a sizeable career / experience, and a strong interest in the game. 


Read my guides there

"I want the game to be just how it was launched, no MM limitations for platoons, some unbalanced match once in a while, and friendly fire should come back as well. The only thing they should remove is statistics." -phony1907, 08 July 2016, 05:45pm - 

I want just that, and replay files.

 






Also tagged with Statistics

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users