Why analysing cards over their overall winrate is wrong

  • Hellcopter's Avatar 260 306 Posts Joined 02/09/2020
    Posted 1 year, 6 months ago

    This post is mostly directed to Nifty but i think it could enlight other people as well and spark some interesting discussion along the way

     

    How tier works: 

    Every deck starts equal at 0% winrate. The decks then face each other and the ones who perform best get to be placed at higher ranks, while the worst ones go to the botton. 
    So the botton of the chain food will be filled with the most unique but terrible ideas: excess of extremely greedy cards, wacky wacky combos, not optimized aggro, Yasuo decks, etc  
    Those "slow" and/or unrefined decks are terrible against any optimized aggressive startegy, even the worst ones; but given enough time they may win sometimes against control decks, even the best ones. 

    So here comes the first big distinction between aggro and control:
    Aggressive strategies will ALWAYS perform better then control against the botton spectrum of the range.
    As long recreational players exist, any current meta isn't relevant at all.


    Why the 50% winrate equals balance is a fallacy:
     

    In a theorical ideal perfect balanced game, every card should have a winrate close to 50%. In the real world that will never happen, even if the game is balanced for several reasons:
    1- Skill gap between players -> The best players know the best time to play each card to increse its effectiveness
    2- Skill gap between cards -> Some cards like Legion Rearguard are easier to play then others The Ruination
    3- Metagame -> Some champions/deck/cards have better mechanics while facing other specific ones, which directly changes the winrate
    4- Synergy -> Some cards perform much better alongside other cards, so those would never see play without their respective synergy
    5- Its boring -> Perfect balance means no further optimiztion for decks. So any change translates into negative winrate.


    Metrics and Winrate by tiers:
     

    We already conclude that even bad aggro strategies perform better at low ranks and they should get less and less effective in the face of more optimized decks along the way.
    For instance, an extremely fast Darius deck will be very good choice to climb lower elos in ladder and it gets more even as it goes up.

    So lets say Darius is balanced sitting at 50% winrate and assume those numbers just an example:
    Platinum + -> 1000 games -> 47% winrate
    Silver/Gold -> 1000 games -> 50% winrate
    Iron/Bronze -> 1000 games -> 53% winrate
    Average winrate / game = 50%

    Now, imagine some time passed, the game gets older and players got smarter enough to understand when to properly play Darius. 
    The metrics now would be something like this:
    Platinum + -> 10 games -> 47% winrate
    Silver/Gold -> 100 games -> 50% winrate
    Iron/Bronze -> 1000 games -> 53% winrate
    Average winrate / game = 52,68%

    Notice that Darius had its winrate increased WITHOUT A SINGLE CHANGE in the game 


    Overall Winrate is a useless stat:
     


    Here is where the overall winrate metric over all tiers together falls appart.

    There are many more players bellow gold then anywhere else, so in short, any agressive card should have a winrate over 50%, and every skill intensive card should have a winrate under 50%. 
    Every single agressive champion like Darius should show an overall winrate over 50% winrate. If an aggressive champiom sits at 50% winrate over all tiers he is not balanced; he is thrash tier. 

    I strongly believe trying to measure cards/decks winrates without taking the tier in which they are played into consideration is extremely stupid and not helpfull to anything outside spoting very obvious imbalances that are either broken or unplayable.

    Hearthstone: Me vs Firebat -> https://www.youtube.com/watch?v=09NCE81owjo

    5
  • KinkyJohnFowler's Avatar 20 2 Posts Joined 05/15/2021
    Posted 1 year, 6 months ago

    Very good write up. I always cringe a little when players try to argue a point using win rates (and play rates) of cards without any context.

    “Why are you complaining about X card? It’s only played in Tier 4 decks!” is a popular one.

     

    1
  • Nifty129's Avatar 390 643 Posts Joined 05/29/2020
    Posted 1 year, 6 months ago

    I'll be honest this is so heavily reliant on jargan to be entirely uninformative. If you have a point make it, don't just slam a bunch of numbers and pretend you're bright...this ain't a measuring contest son.

    Instead I'm gonna show you a clip from Graplr where he uses one of the worst champions according to swim in a TF Zillian list a concept I actually pioneered before him.

    https://youtu.be/YUqjgKoHZ3I

    So when I go to bat for these bad champions and say that everyone is just building them wrong thus far you'll believe me not because I slammed numbers but because I speak of objective truths and reality. Or I actually source my #### to put it more simple.

    -3
  • Leave a Comment

    You must be signed in to leave a comment. Sign in here.