COH2 Skill charts + all-time statistics

Hello!

I was playing around with the new leaderboard and I found "hidden" information. With the new information, it is possible to create variables that measure player's skill. (And yes, the variables are only directional.)

The "hidden" data isn't very interesting, but it explains at least to me, why balance feels so screwed and why automatch is bad. The data can be also used to compare the difficulty of the factions and also to see all-time win-ratios, which are btw. almost perfectly balanced. Weird, huh?


Here are the variables that I managed to find / create:

• Win-Ratio: Win / Lose ratio (Old information)
  
• Squad Survive: What percentage of squads have survived to the end of the match.
  
• Squad K/D Ratio: Kill / death ratio of squads
  
• Vehicle Survive: What percentage of vehicles have survived to the end of the match.
  
• Vehicle K/D Ratio: Kill / death ratio of vehicles
  
• Entity Reinforced: What percentage of dead entities have been reinforced.
  
• Vehicle K/D Ratio: Kill / death ratio of vehicles
  

So, I made a small script that collects and combines the hidden data from (max) top 26 000 players per faction. If a player has played under 10 games with a faction, his / her information cannot be seen.
With the data, we can draw charts that show how the best players differ from bad ones.

Here is soviet 1v1 the average stats from the best players to the whole Soviet player-base. (Higher value means more skill in general.)






We can see that all variables drop down at the very start and then they slowly decrease. If we assume that those variables measure the skill level at least somehow, we can conclude that the required skill to rank up from 200 to 50 is much higher than ranking up from 10 000 to 2 000. In other words, this game has few skilled players and thousands of newbie ones. To understand this chart better, we can use the logarithm table.



We can see that the skill level increases massively, if a player has better ranking than 50. If we take, for instance, average Squad K/D Ratio from the best player and the worst one, our value is around as high as top 50 players have. So, the skill difference is basically massive, which leads to many problems. So, no wonder why automatch cannot create balanced teams.

The difference also explains why this game is so hard to balance. For example, top players lose their squads round 1.75 times less frequently than the bad ones. What does it mean? For the worst players, the squads feels 1.75 times more expensive compared to the others. So, the expensive units are harder to get and they don't give as much reward as they would give if the better player used it. This is not a real problem, since the game should be balanced for the good players.

The problem is that there are few "good" players. If the game was well balanced for the top 200 players, the top 50 would still feel the game as unbalanced, because the skill level increases so radically. If the game was balanced for top 50 players, 99.5 % would feel the game as unbalanced, which is not an ideal situation for Relic.

I don't want to make this weird post any longer, so here is the link to the spreadsheet which includes thousands of numbers and 36 charts. All factions and different game modes :
https://docs.google.com/spreadsheets/d/13pNvW3FD3kByufeGnvG8ObyYfRzIyiiYsRx-mXY9AEk/edit?usp=sharing

One more thing, it is pretty interesting to compare data between factions

Awesome work!

Very cool to see how balanced 1v1 is overall - even at the lower skill levels.

RIP 4v4s Allied win rate is so low there. Although I'd argue that that has more to do with AT vs RT and map than anything else. Axis is probably slightly favored in 4s but not by too much.

very interesting. especially how its only soviet having that 'jump'.