This chart is simply an average of every pro player’s rank in each matchup subtracted from their rank in their mirror matchup. It tells us whether a race overperforms or underperforms by subtracting skill from the equation. This is thanks to the fact that mirror matchups don’t have any imbalance, so we can measure their raw skill with no imbalance. If players do better in ZvP than they do in ZvZ, it means ZvP is zerg favored. So we can subtract their mirror rank from their matchup rank and average that for every pro player, and this gives us the average of how each race does without skill biasing the results.
This data comes from every pro player who has at least 50 recorded games (in each matchup) and who has played at least 1 game in 2020, which is a total of 519 players with tens of thousands of games being used to calculate their ranking (which is highly accurate).
From this we can see that TvP is slightly Protoss favored, and PvZ and TvZ are both highly zerg disfavored. Much to my surprise, Zerg is struggling more against Protoss than against Terran.
Protoss is by far the best race on average, since it’s the only race with all it’s matchups above 0 (meaning that Protoss do better in PvT/PvZ than they do in PvP).
EDIT:
Despite fear of making the analysis any more complex, here are charts that definitively prove mirror performance is a near-perfect indicator of skill.
Skill distribution is the same regardless of race played: https://i.imgur.com/BfYgDVk.png
Mirror performance predicts non-mirror performance on an individual basis with ~90% accuracy; predicts race performance almost perfectly (as an average):
If the data said zerg was overpowered I would gladly state as much just for the satisfaction of proving you wrong. Unfortunately that’s simply not been the case and thus I have to endure endless personal attacks from you.
Calling out your bias isn’t personal. This new dogma you’ve promoted is so you play the victim card and hope blizzard sees people who confront your bias as “personal attacks”. It’s a cute new tactic to get rid of forum members you dislike confronting, but its a slimy one.
The satisfaction of proving all your personal attacks wrong would be worth it in and of itself. You are incapable of seeing SC2 through non partisan glasses. Your tribal mentality is so severe you assume everyone else thinks the same way.
I’m sure someone here will, you will deny, and the circle will start over again. You’re telling me, that in the all the years this game has been out, you’ve never seen a time when zerg was OP? It’s ludicrous.
Aka, you can’t, because you aren’t using evidence and thus the attack is not about the argument but rather is personal. You are ignoring the argument, attacking me, and that is the definition of an ad hominem fallacy:
ad hominem: Attacking a person’s character or motivations rather than a position or argument.
Where? List it in this thread. Show me. Where in this thread do you even make an attempt at having an honest discussion?
I’ve posted raw data that you dislike and for that you attack me as biased. That is your entire contribution to this thread. You dislike the data and you have no argument so you go right to slinging mud in literally your first post in the topic.
Did he mention the data even once? Not that I can see! He probably did not look at graph either… but still call it biased… for proof he says someone will find proof for him! He speaks much nonsense.
Poster does personal attack in every thread on bnet. But especially in your thread. And then they say you are crying victim after they harass again and again!
It is a sight I have not seen in the entire internet. It remind me of the abusive husband who tells to his wife that she deserves it…
Elo measured performance relative to the other players in that specific matchup, it doesn’t inherently measure skill. Comparing ELO ratings in different matchups is grasping at straws.
You seriously need to try to understand how pairwise comparison models work before you start making these terrible analysis.
The Elo rating system is a method for calculating the relative skill levels of players in zero-sum games such as chess. It is named after its creator Arpad Elo, a Hungarian-American physics professor.