Jump to content
Vicroline

Is player performance influencing the win rate?

Recommended Posts

Since this is my first post on the Wotlabs forums I do like to introduce myself. I am a average WoT player with some very basic statistic and R skills and like many other WoT players out there I was somewhat superstitious regarding the randomness of matchmaking. I was wondering why there is no 20 percent win rate or a 90 percent if everything is random (normal distribution doh!)?  From what I have noticed the public opinion of the WoT community suggests that matchmaking is either rigged or entirely random. So I have done some research and stumbled on this post.


Swoopie and Sharpneli from the WoT forums are outlining a matchmaking model based on randomness while holding player performance at a constant level. This model incorporates two teams with 15 players on each team. Each player’s performance is randomly generated except one that is representing the player. In other words Team A has 14 randomly generated numbers that can vary from 0 to 100 plus a fixed value for the player and Team B has 15 randomly generated numbers that can vary from 0 to 100. To decide which team has won a battle the calculated values are summed up and compared with each other. Let’s assume that each player of Team A has a performance value of 50 (14*50) and the player value is 75, while each player of Team B has a performance value of 50 (15*50). If we compare the two sums we get 775 for Team A and 750 for Team B. This means that Team A would have won this round based on the above average performance of our player. I calculated this model in R for 10.000 random battles incorporating three fixed player values (0, 50,100) in three models and this is the result.  

sEEsL81.pngCyKwuYu.pngXC0V93j.png

So what does this tell us? Well as you can see individual player performance does influence the outcome in a random environment. While the player in the first model which is not contributing anything to the battle has a 36,7 % win ratio, the player that is contributing the maximum amount possible has a win ratio of 62,8 %. In addition it tells us that a win rate significant below of our first model and above our third model for an individual player in an random environment is highly unlikely. In other words you can contribute nothing and still would win one in three matches and you could outperform every other player on your team and still lose every third.

Since this is the Wotlabs math section I have included the boxplot of the dependent outcome variable. The horizontal black line is the median and splits the distribution in two equal parts as you can see the line in the first model (0 player performance) is below the zero line wich means that the player in this model has more losses than wins, the rest is self explanatory.

eg8rdK9.png

*Update 1*

I may not have expressed myself clearly enough, my calculations neither prove nor disprove that WoT matchmaking is random. It rather emphasizes that in an purely random environment the performance of an individual player has an impact on the overall win rate given the sample size is large enough. This data represents 10.000 calculated battles for each model, each battle comprises 29 random values reaching from 0 to 100 and one fixed value for the players performance. This is not actual data from Wargaming it is purely generated data. The only insight it gives us is how a random environment reacts to a constant performance. There is however some similarity to the win rate of the WoT playerbase. In the future we could run statistical tests with real values and compare it with a random model. This could tell us if the real world data resembles a random system or not.  

Share this post


Link to post
Share on other sites
6 hours ago, Vicroline said:

Since this is my first post on the Wotlabs forums I do like to introduce myself. I am a average WoT player with some very basic statistic and R skills and like many other WoT players out there I was somewhat superstitious regarding the randomness of matchmaking. I was wondering why there is no 20 percent win rate or a 90 percent if everything is random (normal distribution doh!)?  From what I have noticed the public opinion of the WoT community suggests that matchmaking is either rigged or entirely random. So I have done some research and stumbled on this post.


Swoopie and Sharpneli from the WoT forums are outlining a matchmaking model based on randomness while holding player performance at a constant level. This model incorporates two teams with 15 players on each team. Each player’s performance is randomly generated except one that is representing the player. In other words Team A has 14 randomly generated numbers that can vary from 0 to 100 plus a fixed value for the player and Team B has 15 randomly generated numbers that can vary from 0 to 100. To decide which team has won a battle the calculated values are summed up and compared with each other. Let’s assume that each player of Team A has a performance value of 50 (14*50) and the player value is 75, while each player of Team B has a performance value of 50 (15*50). If we compare the two sums we get 775 for Team A and 750 for Team B. This means that Team A would have won this round based on the above average performance of our player. I calculated this model in R for 10.000 random battles incorporating three fixed player values (0, 50,100) in three models and this is the result.  

sEEsL81.pngCyKwuYu.pngXC0V93j.png

So what does this tell us? Well as you can see individual player performance does influence the outcome in a random environment. While the player in the first model which is not contributing anything to the battle has a 36,7 % win ratio, the player that is contributing the maximum amount possible has a win ratio of 62,8 %. In addition it tells us that a win rate significant below of our first model and above our third model for an individual player in an random environment is highly unlikely. In other words you can contribute nothing and still would win one in three matches and you could outperform every other player on your team and still lose every third.

I see what you mean, and it is pretty clear your performance 100% influences the battle, otherwise everyone would have around 49-50% wins (bc of draws it wouldn't be 50%).

I'm a little unsure about you toting numbers around as fact, when you haven't defined what average player performance is (and it isn't roughly hp in damage as there is some hp left on the battlefield in almost every single game). maximum performance isn't really a thing either, as players like x3n4 (and gasai yuno in one tank) massively go over that *100% performance* mark. the general player performance should be a skewed standard deviation curve if i'm not mistaken, and so the *average game* is going to be vastly different to the settings prescribed. there are simply too many variables to be included in a simple model like that, and therefore i guess its only really useful for drawing general conclusions.

however i guess the use of numbers makes it look important and so pubbies might listen:gachi:

Hehe jokes they are pubbies (all the clans i play with call crown and pyro hackers whenever they are outplayed because they went island on mines in strongholds)

 

But anyways i feel like i didn't make my point well and i bumbled around without really knowing what to say (I'm 16 and therefore don't work with statistics much) but just giving the average player a random data score doesnt really work unless you actually pull it off a graph of average player performance- and therefore while the numbers might be similar, I don't think you should go around portraying these numbers a fact i guess....

But its still pretty sick that there are some hard models showing idiots that if they got their crap together and got gud@tonks they could win games.

Oh yeah i forgot *player performance* is a general concept that is intangible, wn8 and things like that can only go so far to cover every single decision a player makes in a game that influences victory.

Just my 5 bucks (this is a lot more than two cents XD)

Share this post


Link to post
Share on other sites

There's always been this disconnect in peoples heads about single battle/short term variance and the 1000 game averages, this is sort of a nice example as to why having a bad team in a few games means nothing in the long run, effectively 29 people are random noise evenly distributed and the only influencing factor in your solo game WR and DPG iis your input. Everyone gets the same random variance, bad teams, bad rng, lucky shots against them, so none of that actually matters and is not an excuse to hide behind.

Share this post


Link to post
Share on other sites

Ahh the 40:40:20 theory - now with added graphs.

The key element that then further lowers the range of observed win rates are the lower influence ranks / tiers. Hence even the shittest of the shit rarely reach 37% except if only playing high influence tanks.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...