Jump to content


Five Months of WoT Stat Collection - 2k+ Games Collected


  • Please log in to reply
77 replies to this topic

NeatoMan #61 Posted Aug 10 2018 - 03:32

    Major

  • Players
  • 26212 battles
  • 17,985
  • Member since:
    06-28-2011

View PostUnicorn_, on Aug 09 2018 - 20:10, said:

Shills don't like data or facts.

Not sure what your point is.  I don't see anyone complaining about this data.  It is what it is.  It's not really revealing anything i haven't seen in my data, or anything that should get anybody's panties in a bunch.

 

Seems to me it's the riggers and tin toilets that always avoid discussing data and facts.  Just look how adamantly daRock refuses to discuss his own data.  They always refuse to answer questions or divulge all their facts.  Notice how the OP has been forthcoming with every request.  That's how it's supposed to be done.  Nothing wrong here



Pipinghot #62 Posted Aug 10 2018 - 09:06

    Major

  • Players
  • 25960 battles
  • 8,845
  • [IOC] IOC
  • Member since:
    11-20-2011

View PostUnicorn_, on Aug 09 2018 - 20:10, said:

Shills don't like data or facts.

Boneheads who can never win an argument with data or facts accuse people of being shills.



da_Rock002 #63 Posted Aug 10 2018 - 11:19

    Major

  • -Players-
  • 7698 battles
  • 2,986
  • Member since:
    11-24-2016

View PostPipinghot, on Aug 10 2018 - 03:06, said:

Boneheads who can never win an argument with data or facts accuse people of being shills.

 

 

AND the shills respond



BurglarOfBanff_ff #64 Posted Aug 10 2018 - 13:22

    Captain

  • -Players-
  • 25777 battles
  • 1,371
  • [DHO4] DHO4
  • Member since:
    05-04-2014

View PostTrauglodyte, on Aug 08 2018 - 20:51, said:

 

As to the percentage of people spitting out 0 damage games, I'm included in those numbers.  Today, out of 34 games that I played - I'm on vacation, btw, so that explains the larger sample for treason, every three weeks or so, I have a hiccup.  Look at my wotlabs.net data ( http://wotlabs.net/na/player/Trauglodyte ;) and view the win rate graph.  It shows a graph of around 2,000 games and it climbs and then plateaus, climbs and then plateaus.  Right now, I'm on a 2 week bender where I can't break 50%.  Did I forget how to play?  My Average Damage, wn8, and Average Experience per game says that I haven't.  Yet, in the data, I'll go strong for weeks and then *pow* in the face!

 

Same exact thing happens to me.  My graph looks almost the same.  I've always thought that this was caused by MM - not sure why or how.  No, I don't think MM picks on me, we all get the same MM, but there does seem to be some sort of pattern.

 

 



Pipinghot #65 Posted Aug 10 2018 - 19:52

    Major

  • Players
  • 25960 battles
  • 8,845
  • [IOC] IOC
  • Member since:
    11-20-2011

View Postda_Rock002, on Aug 10 2018 - 05:19, said:

AND the shills respond

So how much are WT and AW paying you to post garbage on these forums? Someone who is as intimately familiar with shilling as you claim to be must have experience being one.



Nunya_000 #66 Posted Aug 10 2018 - 19:56

    Major

  • Players
  • 21063 battles
  • 13,040
  • [PACNW] PACNW
  • Member since:
    09-20-2013

View PostPipinghot, on Aug 10 2018 - 00:06, said:

Boneheads who can never win an argument with data or facts accuse people of being shills.

 

View Postda_Rock002, on Aug 10 2018 - 02:19, said:

AND the shills respond

 

 

I think you might have just proved his point.

 

 



Siege_Engine #67 Posted Aug 12 2018 - 04:16

    Captain

  • -Players-
  • 10298 battles
  • 1,158
  • [AP-CR] AP-CR
  • Member since:
    01-26-2015

View PostNeatoMan, on Aug 09 2018 - 13:48, said:

Remember he ommitted games that were won by cap or were draws, which was about 10% of his games.  You can reduce that by 3%.  High 20s is what I get in the upper tiers. Blowouts always increased for me as battle tier goes up, and a lot of his games were high battle tier.  I'd like to get his data to do that kind of comparison.  I'm sure it won't be any trouble, unlike daRock.  Who knows what he's got scribbled on his desk.

 

A time ago you mentioned that I won't reply to your posts.  Here's a good example of why.  

 

You debate +/- 0.5 decimal places.

 

You miss the point, or troll the point.  I don't know which, but you don't move the discussion forward.  



Badabingg #68 Posted Aug 12 2018 - 16:09

    Staff sergeant

  • -Players-
  • 6904 battles
  • 311
  • [W-UN2] W-UN2
  • Member since:
    12-07-2015

"Out of every one hundred men, ten shouldn't even be there, eighty are just targets, nine are the real fighters, and we are lucky to have them, for they make the battle. Ah, but the one, one is a warrior, and he will bring the others back."
    -- Heraclitus

 and this from a guy who had to do his math on a clay tablet

 

The plural of anecdote is not data.



NeatoMan #69 Posted Aug 13 2018 - 14:13

    Major

  • Players
  • 26212 battles
  • 17,985
  • Member since:
    06-28-2011

View PostSiege_Engine, on Aug 11 2018 - 22:16, said:

A time ago you mentioned that I won't reply to your posts.  Here's a good example of why.  

 

You debate +/- 0.5 decimal places.

 

You miss the point, or troll the point.  I don't know which, but you don't move the discussion forward.  

All i'm saying is that his results are not far off from what I got.  It's been this way for a long time.  There is nothing out of the ordinary to be getting 25% to 30% blowouts in the tiers he played.  It has nothing to do with too many stacked teams.  It's simply the nature of the game. 

 

The more people collect and show their data, the more it seems to confirm the numbers I've been getting, and those numbers showed that balancing teams will not significantly reduce blowouts.

 

If daRock would grow a pair and show his data I bet it would fall in line with our two data sets too, but he's obviously not interested in an honest discussion.   Of course after hiding it for so long one begins to doubt the validity of anything he does provide.  He just ends up being another Shadora.

 

Plus, the way data gets twisted around here,  "one third" soon becomes "40% of all battles", which then gets reported as "nearly half of all battles..." by those trying to push their agenda.  Just keeping it honest



Urabouttudie #70 Posted Aug 13 2018 - 14:36

    Captain

  • Players
  • 21043 battles
  • 1,805
  • Member since:
    11-11-2013
Controlled Opposition....

NeatoMan #71 Posted Aug 17 2018 - 02:34

    Major

  • Players
  • 26212 battles
  • 17,985
  • Member since:
    06-28-2011

Got some data from a volunteer using the WotNumbers site.  Trying to get more off that site (they have a few million battle results uploaded).  Should give a good representation of the actual blowout rates in WoT

 

Here is how his results stack up against mine (his are first, mine second).  Pretty much the same as I got.  Still waiting on raw data from the OP, but from what he presented it looks close too.  This is all data from the new MM.

 



SporkBoy #72 Posted Aug 17 2018 - 15:37

    First lieutenant

  • Players
  • 37329 battles
  • 865
  • [PZB] PZB
  • Member since:
    02-06-2014
Piles of data and not one actual hypothesis test with statistical power assessment. You know, basic statistical analysis stuff.

bocaPG #73 Posted Aug 17 2018 - 15:45

    Sergeant

  • -Players-
  • 17432 battles
  • 139
  • [SRB] SRB
  • Member since:
    09-13-2014

Great job OP.

It aligns with my observation that you are screwed in tier 8, disproportional number of low tier games... Not too fun in medium when you can't pen from the side. That tier 7 is gold with the right tank.

Well done.

 



Prosqtor #74 Posted Aug 17 2018 - 16:18

    Major

  • Players
  • 59436 battles
  • 2,013
  • [ONION] ONION
  • Member since:
    12-16-2011
Usually most players do some damage.  However, I DID have a game the other day where 11 players did zero damage.  It was a combination of a great opponent, really bad players, and REALLY bad RNG with regard to pen.  Several of the no damage players LANDED shots.  It was funny in a sick sort of way.

NeatoMan #75 Posted Aug 17 2018 - 16:29

    Major

  • Players
  • 26212 battles
  • 17,985
  • Member since:
    06-28-2011

View PostSporkBoy, on Aug 17 2018 - 09:37, said:

Piles of data and not one actual hypothesis test with statistical power assessment. You know, basic statistical analysis stuff.

It's not trying to test any hypothesis.  It's just a presentation of data answering the question:  How often do blowouts occur?   With regard to this kind of analysis the main request is: more data is better.

 

btw, what hypothesis would you want to test?



sleeper_agent #76 Posted Aug 17 2018 - 18:08

    Major

  • -Players-
  • 21916 battles
  • 2,430
  • Member since:
    06-19-2013
wow. solid work and stats

Trauglodyte #77 Posted Aug 17 2018 - 20:55

    Major

  • -Players-
  • 16092 battles
  • 2,561
  • [GSH] GSH
  • Member since:
    06-04-2016

View PostNeatoMan, on Aug 17 2018 - 02:34, said:

Got some data from a volunteer using the WotNumbers site.  Trying to get more off that site (they have a few million battle results uploaded).  Should give a good representation of the actual blowout rates in WoT

 

Here is how his results stack up against mine (his are first, mine second).  Pretty much the same as I got.  Still waiting on raw data from the OP, but from what he presented it looks close too.  This is all data from the new MM.

 

 

Neato,

 

I saw your IM.  I've been on vacation and will get you my data set when I get back.

 

As to the one poster's comment on hypothesis, it is really hard to derive something in which to test because you don't have a control group nor do we have the ability to pull levers within the game to see the impact of changes.  All that I have access to is my personal performance data and the end game data of my teammates.  To really do a big statistical deep dive, I'd need a monster amount of data because there is a lot that isn't included in my original findings.  Case in point, I track my average damage per game per tank and compare that against the average of the team, to see where I'm performing against my peers.  The problem is that damage numbers are largely impacted by the length of a game, the map upon which the game is being played, and the quality of people with whom I'm playing.  Plus, while I'm the only constant in my data set, I'm not always the same player from game to game because my performance is predicated upon the choices I make based upon the environment of the game on that map at that point of the game given the resources that I have (health, ammo, teammates, etc.) and what I'm facing.  I could play my Panther 88 one million times on Mountain Pass, all starting from the same spawn point, and have different results across all one million games because the variables change from game to game and from moment to moment.

 

I'd love to hookup with the WarGaming folks and do some crunching or, at the very least, get a peak into how they gather and crunch data.  Even if I had to sign an NDA, it would extremely intriguing because of how everything differs.  We complain about some tanks being bad but I often wonder if tanks are really bad or if there is a greater population of bad players playing specific tanks which, in turn, drives personal feelings about tanks.  The Tiger II, case in point, is not a great tank but it isn't bad.  But, I've seen a ton of turds play it at turd levels.  Does that mean that the hypothesis of tank buffing correlates with the tank?  Or, does it mean that the tank has a higher skill cap and that people are incapable of playing it at a high level given that cap?  Is there Russian bias or do Russian tanks require less skill but produce a higher return on skill investment?  You could go around and around in circles and never get anything concrete out of it.  But, I'd love to give it a go.



NeatoMan #78 Posted Aug 17 2018 - 21:13

    Major

  • Players
  • 26212 battles
  • 17,985
  • Member since:
    06-28-2011

But if you get a number of people who all get similar results, then you can be pretty sure that's the norm. We probaby can dismiss claims of "I get blowouts half the time" or "I'm getting way more blowouts this patch", without having to analyze a huge amount of data.

As far as tank win rates and player quality that can be addressed by tank curves found on the wotnews site

http://ftr.wot-news....nk-performance/


Edited by NeatoMan, Aug 18 2018 - 00:23.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users