Jump to content


Statistical analysis: Expected number of win/loss streaks

Statistics Analysis Streaks Simulation

  • Please log in to reply
49 replies to this topic

gallabru #41 Posted 21 January 2019 - 10:38 AM

    Senior Sergeant

  • Player
  • 17980 battles
  • 641
  • [LAGER]
  • Member since:
    09-01-2016

You all are right : there are most likely external (i.e. player related factors) that affect the distribution of wins and losses compared to a purely random one. Playing golf, I know that the most difficult thing to do is forgetting the previous shot. 

The thing is : we don't know what is the current distribution and how much it departs from a random one, apart from the 'guts proofs' proposed so far. 

Once we have it, we will be able to discuss instead of comparing statements. 


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

Carlsberg clan family is recruiting. Contact us.


Jukkis74 #42 Posted 21 January 2019 - 04:20 PM

    Senior Sergeant

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

That academic article has interesting bit included relatex to our earlier discussions - just by observing random sequence of 200 coin tosses people started to perceive what random is better than they did earlier!

 

I swear I did not try to be academic when I suggested somewhere in MM thread that people should write down the results of 100 (or so) coin tosses :-)



Jukkis74 #43 Posted 21 January 2019 - 04:23 PM

    Senior Sergeant

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

View Postgallabru, on 21 January 2019 - 12:38 PM, said:

You all are right : there are most likely external (i.e. player related factors) that affect the distribution of wins and losses compared to a purely random one. Playing golf, I know that the most difficult thing to do is forgetting the previous shot. 

The thing is : we don't know what is the current distribution and how much it departs from a random one, apart from the 'guts proofs' proposed so far. 

Once we have it, we will be able to discuss instead of comparing statements. 

 

Once we get going, we can try to estimate player-specific ’flow’ and ’tilt’ factors, i.e. if some players are prone to really good or bad series. I know some players (better than myself) that I would guess have quite high tilt factor. 

jylpah #44 Posted 22 January 2019 - 11:04 PM

    First Sergeant

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

View PostPururut, on 21 January 2019 - 12:24 PM, said:

I think I dont understand. Is the data from analyzing matchmaker which is random by its very nature reliable? Even if it is there are so many non-mathematical factors involved ranging from the stacking moral gain/loss, players willingness to take the game seriously, communicate and play as a team, etc. I appreciate the work but dont know whether if it is suited to analyzing the matter involved.

 

I think I don't understand. :amazed:

 

The whole point of the test is to compare the MM/win/loss streak to truly random data. I do not think I have patience to collect enough data though. It might be that to make statistically significant findings one needs quite a many battles.

 

Anyway, when tanking feels boring, the math can be fun.


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

Jylpah @EU  Jylpah @Asia  Blitz-tools  


sixty_three #45 Posted 23 January 2019 - 09:30 AM

    First Sergeant

  • Player
  • 12321 battles
  • 1,343
  • Member since:
    04-29-2017

Loss streaks of 3 or more are known to increase the rate at which new players quit playing a game.  I could imagine that WG would like to tinker around with the MM to reduce the likelihood of such streaks occurring.

 

Tilt and flow effects on the other hand would tend to increase the likelihood of longer win and loss streaks. 

 

I don’t see how you could prove that any statistically significant variance against random win/loss forecasts was down to, say 75% MM tweaking and 25% player factors.  But you could probably make an educated hunch.

 

Finding nothing would also be interesting.  


One of life’s challenges — knowing enough to think you are right, but not enough to know that you are wrong.

 


Jukkis74 #46 Posted 23 January 2019 - 08:12 PM

    Senior Sergeant

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

View Postsixty_three, on 23 January 2019 - 11:30 AM, said:

Loss streaks of 3 or more are known to increase the rate at which new players quit playing a game.  I could imagine that WG would like to tinker around with the MM to reduce the likelihood of such streaks occurring.

 

Tilt and flow effects on the other hand would tend to increase the likelihood of longer win and loss streaks. 

 

I don’t see how you could prove that any statistically significant variance against random win/loss forecasts was down to, say 75% MM tweaking and 25% player factors.  But you could probably make an educated hunch.

 

Finding nothing would also be interesting.  

 

Tilt, flow, tilt-bounce and flow-reverse. Better terms welcome for the last two :-) 

 

Tilt-bounce would be the increased likelihood to get a win after several losses. Flow-reverse increased likelihood to lose after consecutive wins.



Replicant_2 #47 Posted 23 January 2019 - 10:43 PM

    First Sergeant

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

This conversation about probability of winning/losing depending on the preceding streak reminded me of sth I heard on the radio a long time ago.

Not tank related but an equally entertaining statistical analysis:

 

https://www.cambridg...64E3DA5696DA555


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

Time to die.


gallabru #48 Posted 24 January 2019 - 01:07 PM

    Senior Sergeant

  • Player
  • 17980 battles
  • 641
  • [LAGER]
  • Member since:
    09-01-2016

I have started some analysis on my alt account using Jylpah's codes and Mjr_Eazy's hotline.

I have 2700 battles on this one. Over 15 battles here is what I get :

TOTAL____________ Battles WR DPB DR KDR Spot Hit rate Pen rate Surv% T alive% Top tier Allies WR Enemies WR Allies Btls Enemies Btls
  15 86.7% 1393 1.7 6.0 1.3 84.0% 89.3% 73.3% 94% 67% 49.14% 47.61% 2650 2846

WR is meaningless on such small sample. I have started playing the amx 50 120, so for 3 battles I was necessarily top tier. My rate of top tier would be 58% without these battles. I suppose that the WR of the allies is also a bit larger because of my contribution to this value. My average tier played was 7 (weighted for tiers ranging from 5 to 9). 

I am trying to see if I can get the data of all players, but my general feeling is shown here : I am seeing only players below 5k battles. I know that when queueing time is long, I will be dropped in a battle with players with >5k battles. This is more exception than generality.

 


Edited by gallabru, 24 January 2019 - 01:08 PM.

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

Carlsberg clan family is recruiting. Contact us.


jylpah #49 Posted 27 January 2019 - 07:04 PM

    First Sergeant

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

View Postgallabru, on 24 January 2019 - 03:07 PM, said:

I have started some analysis on my alt account using Jylpah's codes and Mjr_Eazy's hotline.

I have 2700 battles on this one. Over 15 battles here is what I get :

TOTAL____________ Battles WR DPB DR KDR Spot Hit rate Pen rate Surv% T alive% Top tier Allies WR Enemies WR Allies Btls Enemies Btls
  15 86.7% 1393 1.7 6.0 1.3 84.0% 89.3% 73.3% 94% 67% 49.14% 47.61% 2650 2846

WR is meaningless on such small sample. I have started playing the amx 50 120, so for 3 battles I was necessarily top tier. My rate of top tier would be 58% without these battles. I suppose that the WR of the allies is also a bit larger because of my contribution to this value. My average tier played was 7 (weighted for tiers ranging from 5 to 9). 

I am trying to see if I can get the data of all players, but my general feeling is shown here : I am seeing only players below 5k battles. I know that when queueing time is long, I will be dropped in a battle with players with >5k battles. This is more exception than generality.

 

 

If you could run the script again for 100+ battles that would be great.

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

Jylpah @EU  Jylpah @Asia  Blitz-tools  


gallabru #50 Posted 27 January 2019 - 08:30 PM

    Senior Sergeant

  • Player
  • 17980 battles
  • 641
  • [LAGER]
  • Member since:
    09-01-2016
I am on my way to it...

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

Carlsberg clan family is recruiting. Contact us.






Also tagged with Statistics, Analysis, Streaks, Simulation

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users