Jump to content


How player skill affects the WR?

statistics work numbers WR skill

  • Please log in to reply
94 replies to this topic

Enforcer1975 #21 Posted 19 April 2017 - 03:41 PM

    Lieutenant General

  • Player
  • 12716 battles
  • 7,912
  • [WJDE] WJDE
  • Member since:
    05-04-2014

View PostUllakkomorko, on 19 April 2017 - 02:49 PM, said:

 

You could have a team chess game where every player takes turns to move a piece without having the chance to communicate with each other apart from being able to call the players on your own team noobs and worse! Imagine the frustration of the unicum grandmaster in a team of reds beginners.

But a half decent chess player would still make the right move more often than the tomato bob in WoT. And you only move a piece per turn unlike in the game where ( almost ) everyone is on the move. Almost all of those who know more can take the initiative and those who can't can only react to what happens often to their disadvantage because they can only see limited options for the situation they are in. 


Edited by Enforcer1975, 19 April 2017 - 03:42 PM.


Str0nkTenk #22 Posted 20 April 2017 - 12:37 PM

    Warrant Officer

  • Player
  • 37912 battles
  • 576
  • Member since:
    12-19-2012

so lets crack on with the chess analogy..

 

so we're saying 15 v 15 chess players, where each one gets to do their own thing, but the referee gets to pick the teams and which piece they can play.

 

what would be the outcome of the situation where 4 grandmasters and 11 average joes with pawns as pieces go against a whole 15 noob team of queens, on a board that suits queen gameplay.

 

coz thats WG balancing, right there.

 

at least in chess, there is a close to 50% chance for each player to win, at the start, with both players starting with the same tanks, on an even playing field.

in WOT, there is not.

 

and also, in the real world, grandmasters do not play against noobs. 

noobs play noobs,

grandmasters play grandmasters,

and average joes play against average joes.

in that way, everyone has fun, regardless of skill, and everyone has an even chance to win at the start of the game

 

 

 


Edited by Str0nkTenk, 20 April 2017 - 12:38 PM.


shishx_the_animal #23 Posted 20 April 2017 - 12:40 PM

    Lieutenant Сolonel

  • WGL PRO Player
  • 28209 battles
  • 3,093
  • [FAME] FAME
  • Member since:
    04-06-2013
That was an extremely poor analogy, sorry mate.

jabster #24 Posted 20 April 2017 - 12:46 PM

    Field Marshal

  • Beta Tester
  • 11854 battles
  • 16,408
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View Postshishx_the_animal, on 20 April 2017 - 11:40 AM, said:

That was an extremely poor analogy, sorry mate.

 

It would have been good if he had said the referee picks on one player while the other twenty nine players are mere pawns for reasons.

Excavatus #25 Posted 20 April 2017 - 12:57 PM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

Ok,

I found some time between work and threw some die,

preciesly for 400 times more,

now the random skill battle numbers are 500.. and here are the results.

 

 

AFTER 500 matches played in random,

 

The average total skill level of the 15 player red team is : 750,45

The average total skill level of the 14 player Gren team is : 698,86

The average for the players A, B, C, D, E and F can be respectively found adding their skill levels, but anyway here they are,

748.86, 758.86, 768.86, 778.86, 788.86 and 798,86

 

 

First of all, total rofflstomps,

The red team won, %37,80 of the matches, regardless the last player we are looking for.

Green Team won, %38,20 of the matches, even they are 14vs15.

 

So, our constant player, cannot and will not change the outcome of his %37,80 losses and cannot make his team loose the %38,20 of the matches.

 

the turnback numbers for our players are,

When we add them into the equation,

 


Player A, turned 55 of the losses into the wins, and he has a total of %49,20 WR by winning 246 out of 500

Player B, turned 66 of the losses into the wins, and he has a total of %51,40 WR by winning 257 out of 500

Player C, turned 81 of the losses into the wins, and he has a total of %54,40 WR by winning 272 out of 500

Player D, turned 99 of the losses into the wins, and he has a total of %58,00 WR by winning 290 out of 500

Player E, turned 105 of the losses into the wins, and he has a total of %59,20 WR by winning 296 out of 500

and finally, our ubersuperultraunicumpurple player F,

Turned 119 of the losses into the wins and he has a total of %62 WR by winning 310 out of 500


 

And the turn rates for the players, (turning losses into the wins, calculated by, (total number of the games that the player turned divided by the total number of the games the player turned plus total number of the games the red team won))

 


A: can turn losses into wins with %17.80 probability

B: can turn losses into wins with %21.36 probability

C: can turn losses into wins with %26,21 probability

D: can turn losses into wins with %31,04 probabilty

E: can turn losses into wins with %33,98 probabilty

F: can turn losses into wins with %38,51 probabilty


Some numbers for people who are interested,

Median of the R team skill : 748

Median of the 14 people G team skill: 695

Standard Deviation of R team for 500 matches: 120,03

Standard Deviation of G (14) for 500 matches: 108,42

We have 1 Draw in 500 matches.. at the initial start and green team wins when the player A joins.

 

Discuss!, what you make of these numbers?

God! I love playing with numbers!


Edited by Excavatus, 20 April 2017 - 01:01 PM.


Ullakkomorko #26 Posted 20 April 2017 - 01:20 PM

    Staff Sergeant

  • Player
  • 11382 battles
  • 359
  • [HKI] HKI
  • Member since:
    04-12-2013

View PostExcavatus, on 20 April 2017 - 12:57 PM, said:

So, our constant player, cannot and will not change the outcome of his %37,80 losses and cannot make his team loose the %38,20 of the matches.

 

This is pretty interesting. If you add those two numbers together, one player could only have an impact on a bit less than 25% of matches. In the remaining matches he could just drive around the map and give an illusion of adding value (so as not to discourage his team mates) and it wouldn't make a difference. This is probably quite close to the truth in WoT. Since its a team game, there are a lot of matches where you can do your utmost best and still lose and a lot of games where you can accidentally drive off a cliff and still win. DIfferences in team composition and the effect of map on different tanks could even decrease the number of games where you can actually make a difference between win and loss.

 

We tend to view matchmaking as black (rigged mm or stupid teams decide everything) or white (if you lose, you did something wrong) whereas the truth is a shade of grey. Your numbers seem to suggest that it's a pretty dark shade of grey.



Excavatus #27 Posted 20 April 2017 - 01:45 PM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

View PostUllakkomorko, on 20 April 2017 - 03:20 PM, said:

 

This is pretty interesting. If you add those two numbers together, one player could only have an impact on a bit less than 25% of matches. In the remaining matches he could just drive around the map and give an illusion of adding value (so as not to discourage his team mates) and it wouldn't make a difference. This is probably quite close to the truth in WoT. Since its a team game, there are a lot of matches where you can do your utmost best and still lose and a lot of games where you can accidentally drive off a cliff and still win. DIfferences in team composition and the effect of map on different tanks could even decrease the number of games where you can actually make a difference between win and loss.

 

We tend to view matchmaking as black (rigged mm or stupid teams decide everything) or white (if you lose, you did something wrong) whereas the truth is a shade of grey. Your numbers seem to suggest that it's a pretty dark shade of grey.

 

You are completely right,

it is not rigged entirely but it is not depended on the skill solely.

Actually I am now on the 800 matches in the simulation, I will post results after hitting the 1000,

My initial target is 20.000 If I dont get bored..

 

When you look at the numbers, yes for the average 50 skill bob, after 500 matches, he lost 254, and won 246 matches,

the 189 of those matches are the red team roflstomps! so he had no affect, %37,80

the 191 of those matches are the green team roflstomps! so he had no affect, %38,20

So he has no affect on %76 of the matches he plays,

then the rest %24, (120) he can actually play a part on whether his team loses or wins..

 

When you look at the numbers, that looks aweful.. %76 of the games that it doesnt matter what you do looks aweful..

but.. as a member of 15 player team, actually your role looks like %6 If everyone contributes enough..

So, changing losses into wins by %17,8 in your games.. I say it is a good number, and skill is a very very good factor here about the outcome of the battle..

 

Just look at the %62 WR unicum player,

He can change the result of %38,51 of his losses..

That means he can directly affect the result of his 100 battles by the percent of %23,8

That is quite good I say..



jabster #28 Posted 20 April 2017 - 01:48 PM

    Field Marshal

  • Beta Tester
  • 11854 battles
  • 16,408
  • [WSAT] WSAT
  • Member since:
    12-30-2010
Honestly recode what you've done to use a local RNG as it will be a lot less painless.

Excavatus #29 Posted 20 April 2017 - 02:30 PM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

View Postjabster, on 20 April 2017 - 03:48 PM, said:

Honestly recode what you've done to use a local RNG as it will be a lot less painless.

 

If I know, I would, but I dont know, I cant! :)

rolling die is much more fun anyway, I am a hardcore retro ADnD DM...



jabster #30 Posted 20 April 2017 - 02:51 PM

    Field Marshal

  • Beta Tester
  • 11854 battles
  • 16,408
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostExcavatus, on 20 April 2017 - 01:30 PM, said:

 

If I know, I would, but I dont know, I cant! :)

rolling die is much more fun anyway, I am a hardcore retro ADnD DM...

 

I'm sure you if you post the bit of code you're using then someone we'll be able to help out.

Kozzy #31 Posted 20 April 2017 - 03:16 PM

    Corporal

  • Player
  • 34172 battles
  • 176
  • [EAB2] EAB2
  • Member since:
    06-29-2011

View PostStr0nkTenk, on 19 April 2017 - 11:42 AM, said:

Great lab maths.

However WOT matchmaking is not 'random'.

Where are the 15 v 15 arty games?

Where are the 15 v 15 light tank games?

Where are the 15 v 15 tier10 heavy games?

The game is completely unbalanced due to the matchmaking. Most games these days are decided before the countdown is ended. Regardless of player skill.

Your random mathematics take equality of chance into consideration.

WG does not apply the same fairness to its matchmaking

End of thread

 

You are (just like many other complete maths rejects) forgetting (denying?) that this magical, mystical MM AFFECTS BOTH TEAMS' PLAYERS not just 'Str0nkTenk' or that other delusional pancake 'QPRanger'.

 

WG was not started as a way of picking on you.  You are not special.  You just find maths more difficult than it actually is.

 

If everyone is affected by the same evil MM then the ONLY constant in EVERY game you play is YOU.  Otherwise you are basically saying that YOU are being victimized by WG (which does make you look like a complete tool).



Ankara_Aatu #32 Posted 20 April 2017 - 03:42 PM

    Staff Sergeant

  • Player
  • 24599 battles
  • 412
  • [RIOD] RIOD
  • Member since:
    04-11-2015

View PostUllakkomorko, on 20 April 2017 - 01:20 PM, said:

 

This is pretty interesting. If you add those two numbers together, one player could only have an impact on a bit less than 25% of matches. In the remaining matches he could just drive around the map and give an illusion of adding value (so as not to discourage his team mates) and it wouldn't make a difference. This is probably quite close to the truth in WoT. Since its a team game, there are a lot of matches where you can do your utmost best and still lose and a lot of games where you can accidentally drive off a cliff and still win. DIfferences in team composition and the effect of map on different tanks could even decrease the number of games where you can actually make a difference between win and loss.

 

We tend to view matchmaking as black (rigged mm or stupid teams decide everything) or white (if you lose, you did something wrong) whereas the truth is a shade of grey. Your numbers seem to suggest that it's a pretty dark shade of grey.

 

If we wish to know the percentage of "unwinnable" (and conversely auto-win)games, we could find out the best solo win ratios on the servers (limited to a reasonably high tier if we like). I think it's safe to assume that in a popular game like WoT the actual top players are fairly close to what is humanly possible. If (playing public matches in) WoT was a professional sport where the athletes are professionally trained from a young age and picked from a huge population, the top players would probably be slightly better, but not by a large absolute margin. So we should conclude that the humanly possible maximum solo win ratio is maybe a few percentage points above the actual empirical maximum found on the servers. If that number is, say 70%, we can then conclude that 30% of games are unwinnable, and by contrast 30% are auto-wins. Which would leave roughly 40% of the games in the grey area where your actions will decide the outcome.

 

Of course from the point of view of an average player the grey area will in practice be much smaller, since he lacks the skills necessary to ever turn the games in the outer reaches of that area.


Edited by Ankara_Aatu, 20 April 2017 - 03:43 PM.


Excavatus #33 Posted 21 April 2017 - 06:44 AM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

View Postjabster, on 20 April 2017 - 04:51 PM, said:

 

I'm sure you if you post the bit of code you're using then someone we'll be able to help out.

 

I am using an excel work sheet :)

 

View PostAnkara_Aatu, on 20 April 2017 - 05:42 PM, said:

 

If we wish to know the percentage of "unwinnable" (and conversely auto-win)games, we could find out the best solo win ratios on the servers (limited to a reasonably high tier if we like). I think it's safe to assume that in a popular game like WoT the actual top players are fairly close to what is humanly possible. If (playing public matches in) WoT was a professional sport where the athletes are professionally trained from a young age and picked from a huge population, the top players would probably be slightly better, but not by a large absolute margin. So we should conclude that the humanly possible maximum solo win ratio is maybe a few percentage points above the actual empirical maximum found on the servers. If that number is, say 70%, we can then conclude that 30% of games are unwinnable, and by contrast 30% are auto-wins. Which would leave roughly 40% of the games in the grey area where your actions will decide the outcome.

 

Of course from the point of view of an average player the grey area will in practice be much smaller, since he lacks the skills necessary to ever turn the games in the outer reaches of that area.

 

actually no, the grey area is all the same for all the players, %35 autowins, %35 autoloss,

then the last %30 is the grey are for all,

Unicums tend to sit closer to the white end, bots to the black,

and average bob sits in the middle..

 

thats all..



balancano24 #34 Posted 21 April 2017 - 07:36 AM

    Corporal

  • Player
  • 3671 battles
  • 136
  • Member since:
    09-13-2016
did you "nerf" dice? if not, all of your hard work is futile. :teethhappy:

Excavatus #35 Posted 21 April 2017 - 11:30 AM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

OK gentelmen,

Now I hit the 1000 games in the silly simulation,

gonna include the "GREY ZONE" term into the results.

First of all results... in a short version,


Number of matches played: 1000

Number of red team roflstomps: 380 / %38

Number of 14-player green team roflstomps: 388 / %38,8

Number of draws: 4 / %0,4


Now lets look at our individual players,

For these 1000 matches


Player A, turned 106 of the losses into the wins, and he has a total of %49,40 WR by winning 494 out of 1000

Player B, turned 129 of the losses into the wins, and he has a total of %51,70 WR by winning 517 out of 1000

Player C, turned 156 of the losses into the wins, and he has a total of %54,40 WR by winning 544 out of 1000

Player D, turned 191 of the losses into the wins, and he has a total of %57,90 WR by winning 579 out of 1000

Player E, turned 204 of the losses into the wins, and he has a total of %59,20 WR by winning 592 out of 1000

and finally, our ubersuperultraunicumpurple player F,

Turned 228 of the losses into the wins and he has a total of %61,60 WR by winning 616 out of 1000


As For the Turn Rates


A: can turn losses into wins with %17.32 probability

B: can turn losses into wins with %21.08 probability

C: can turn losses into wins with %25,46 probability

D: can turn losses into wins with %31,21 probabilty

E: can turn losses into wins with %33,33 probabilty

F: can turn losses into wins with %38,25 probabilty


This is how individual skill can affect the battle results.

But what about the thing called GREY ZONE

 

So clearly, you have a say on the result of only the %23,2 of the battles you played in random.

I guess that is why some of the great players I've known, dont play randoms at all.. after they re-roll :)



Edited by Excavatus, 21 April 2017 - 11:32 AM.


jabster #36 Posted 21 April 2017 - 11:49 AM

    Field Marshal

  • Beta Tester
  • 11854 battles
  • 16,408
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostExcavatus, on 21 April 2017 - 05:44 AM, said:

 

I am using an excel work sheet :)

 

 

Download Python as that's pretty simple to start with and there's quite a few people around that can help.

Ullakkomorko #37 Posted 21 April 2017 - 11:58 AM

    Staff Sergeant

  • Player
  • 11382 battles
  • 359
  • [HKI] HKI
  • Member since:
    04-12-2013

Good job so far! There's a certain flair to doing this one simulated match at a time.

 

I'm still wondering what effects normalising skills towards the middle (by which I mean assuming that average skills are more likely than the extremes) would have. I've never used XVM but I assume that the probability of having two or more super unicums (say 95-100) or utter lossmachines (0-5) are too high in your current system. Obviously the two extremes cancel each other out and tend towards the average but it would give a greater weight our brave individual players skill.

 

I'm not much of a coder myself but I have an add on to Excel that I could build the simulation in. I'll try to summon the will to do that which is hard due to it being Friday afternoon.



Excavatus #38 Posted 21 April 2017 - 12:41 PM

    Second Lieutenant

  • Player
  • 17319 battles
  • 1,325
  • [T-D-U] T-D-U
  • Member since:
    07-28-2013

Actually the macros are set in the excel and I just enter the die results

and I love using the numeric keyboard..

 

On the other hand, I was just thinking about that,

what happens I add a 3xsuper unicum player into a 12-player random team..

how much they influence? how big is their grey zone?

ok.. it is friday, and I have time for that.. now going in for toons now :)



jabster #39 Posted 21 April 2017 - 12:52 PM

    Field Marshal

  • Beta Tester
  • 11854 battles
  • 16,408
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostUllakkomorko, on 21 April 2017 - 10:58 AM, said:

 

I'm still wondering what effects normalising skills towards the middle (by which I mean assuming that average skills are more likely than the extremes) would have. I've never used XVM but I assume that the probability of having two or more super unicums (say 95-100) or utter lossmachines (0-5) are too high in your current system. Obviously the two extremes cancel each other out and tend towards the average but it would give a greater weight our brave individual players skill.

As I said the problem is that in a single match you're trying to model performance for that game not overall skill. Take you average match and then look at the top and bottom three by exp. 



brumbarr #40 Posted 21 April 2017 - 12:55 PM

    Captain

  • Player
  • 32117 battles
  • 2,213
  • [OMNI] OMNI
  • Member since:
    07-30-2012

View PostExcavatus, on 21 April 2017 - 12:41 PM, said:

Actually the macros are set in the excel and I just enter the die results

and I love using the numeric keyboard..

 

On the other hand, I was just thinking about that,

what happens I add a 3xsuper unicum player into a 12-player random team..

how much they influence? how big is their grey zone?

ok.. it is friday, and I have time for that.. now going in for toons now :)

 

Very interesting work. While it gives a great explanation for how WR depends on skill in a random MM. I was wondering if there are any variables you can tweek to get it to agree better with the game. What I mean is that your unicum wins max 62%. Whereas in the game, unicums playing solo usualy win around 65-66% of their games. Is there anything you could adjust for this?





Also tagged with statistics, work, numbers, WR, skill

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users