RE: Usage/win rate of monsters and summoners for each league | Season 49

avatar

You are viewing a single comment's thread:

Interesting post!

I don't think you can directly interpret how good a card is based on its win rate. Or even conclude that cards with a high win rate are causing the high win rates. It might be that a card is expensive and the ability to play it correlates with having an expensive collection. Therefore, a high win rate might just mean that the most fierce players own that card at max level.

The heatmap plots look cool. Did you make them in seaborn / matplotlib? Is the win percentage referring to the win rate of the summoner on the y-axis or x-axis?

hive-13323
0
0
0.000
1 comments
avatar

Thanks and sorry for late reply :)

The heatmap plots look cool. Did you make them in seaborn / matplotlib

Yes, seaborn.heatmap.

Is the win percentage referring to the win rate of the summoner on the y-axis or x-axis?

Oh, I probably should have put a clear info about this. If you want stats for Yodin Zaku, look this way:
image.png

I don't think you can directly interpret how good a card is based on its win rate. (...)

I agree with that! But, I believe that you can still find out a lot of interesting information, especially if you know what you are looking for.
For example:

  • Mylor Crowling is really strong at Bronze level, so it should be on the priority list for any competitive player
  • legendary summoners dominate Champion league
  • which new Dice cards (that are still soo expensive) buy first? Mylor Crowling and Grenadier, then Lorna Shine, Battering Ram etc
  • what about Untamed? Failed summoner, Khmer Princess, Tower Gryffin etc.

Personally, I think that taking into account the top100 players is great for general stats (and better than taking into account all accounts, no value in data from dumb bots or really weak players). And it is useful to decide which cards buy first. But.. if someone wants to improve his gameplay, better look at top10 or just compare his own data with other top player data to see e.g. which cards he underestimates etc.

0
0
0.000
Reply