Context

Clash Royale is a wildly popular game that revolves around fast 1v1 battles. The game is unique in that, along with skill, a player's "deck" (exactly 8 deployable troops the player picks) has also been widely thought to affect one's chances in the game. It is intriguing to be able to see how the endless world of deck configurations match up.

Content

The dataset has two files:

• data_ord.csv: This is juicy part. A collection of 3/4ths of 1 million 1v1 battles recorded from the game.
  • Each row is 1 battle
  • The first 8 columns describe Player 1's deck, representing each card with a number
  • The next 8 columns describe Player 2's deck, representing each card with a number
  • The next 2 columns describe the respective trophy counts of Players 1 and 2
  • The last column is the outcome of the battle; 1 if Player 1 wins, and 0 if Player 2 wins
  • IMPORTANT: Player decks HAVE NO ORDER (In data_ord.csv, they are sorted by ascending card ID, but they can be in any arrangement in real life), and card ID numbers are all categorical
• cardlist.csv: You will notice that rather than the names of a card, they are instead described as integers in data_ord.csv. This file translates each card ID from data_ord.csv into its corresponding card name.

Acknowledgements

Data obtained from the Clash Royale API

Inspiration

I collected this data in hopes to see if the aforementioned "deck configuration" and "trophy counts" could, for the most part, fully predict the outcome of a game.