Low ELO rank as a function of games lost vs terran

General Discussion
1

Some of you might get a kick out of this:

ELORank = round(GamesLostVsTerran ^ IsKr + GamesWonVsTerran + GamesWonVsProtoss + GamesWonVsZerg / (GamesLostVsZerg + GamesLostVsTerran / 12.470112 + 2.650609) + -8.058222 * 23.680236)

This is an equation that quite accurately predicts ELO rankings of pro players based on games won/lost per matchup, and what region that player is from.

Do you see the “GamesLostVsTerran ^ IsKr” term? It’s later divided by GamesLostVsTerran. IsKr is 1 if the player is Korean and 0 if the player is non-Korean. This means that it doesn’t divide by GamesLostVsTerran if that player is a korean. In other words, there is a huge correspondence between losing games vs terran, being non-Korean, and having low ELO.

Here is a chart to demonstrate the concept:

https://i.imgur.com/AyMXU8E.png

This is a slice of the curve with all variables held at mean values except for GamesLostVsTerran, which allows us to see how GamesLostVsTerran controls ELO for the mean-case scenario.

This clearly depicts a trend in the data that non-Koreans have their ELO rankings hurt much more by losing to Terrans. Maybe we need to rethink the idea that in the foreign scene Zerg or Protoss is the easiest race.

1 Like
2

I’m surprised no one is feeding you so let me help. You can have a ladder with 1 terran being better than majority of the non-Terrans. Take heromarine, if he doesn’t play vs SHowtime or Serral he would likely win most of the time -> you get stats that races lose too much from terran -> claim that Terran is (not hardest) - this neither proves it is hardest, nor easiest. You’ve got logical fallacy, same as - maru vs non-Koreans = Terran OP cause 100% winrate right?

Your method of using stats to make a point is :poop:

3

That would make sense if this was a correlation rather than causation. Players factually have low elo as a result of losing to terran players. This shows how much it hurts your ELO to lose to terrans. It’s one of the biggest trends inside the data - losing to non korean terrans NUKES your ELO rank.

This equation accurately models the relationships in the data and it clearly shows that winning or losing against terran is the biggest predictor for ELO success especially in the foreign scene. Whether you win or lose against terran is the biggest factor in determining elo. That clearly paints the picture that having high ELO is difficult due to terran opponents. Ergo, terran is easier.

1 Like
4

Now say this to pro players, I want to see how many laughs you get. I dont want to follow or even get deeper into your stats because theres always something manipulative that can make it look one way

5

Denying a conclusion on the mere possibility that it could be wrong is called circular reasoning and by that metric there is no such thing as truth. Your position is absurd.

That is exactly why I am putting the data out there. I don’t care if a child thinks he is a cat, he is factually not a cat. Whether or not this person or that person thinks terran is hard doesn’t have any meaning compared to quantitative data about reality.

6

If losing vs terran is so punitive, could be cause is the minimum requirement (terran is average weaker)?

7

I gave your thread a heart for the effort alone.

:nerd_face:

8

I think you balance warriors need to take a rest. There are multiple 7k Terran, protoss, and zerg players right now across all the servers. Balance isn’t the reason you’re not 7k.