Lichess Player Performance Dataset

Columns

• game_id: A unique identifier assigned to each individual chess game. There are 20,100 valid and unique game IDs.
• rated: A boolean indicator specifying whether the game was rated (81% true) or unrated (19% false).
• turns: The total number of moves made in a game. The average game length is 60.5 turns, with games ranging from 1 to 349 turns.
• victory_status: Describes how the game concluded, including 'Resign' (56%), 'Mate' (32%), and 'Other' (13%). There are 4 unique statuses.
• winner: Identifies the winner of the game. White won 50% of games, Black won 45%, and 5% had another outcome (e.g., draw). There are 3 unique winner types.
• time_increment: Details the time control setting for the game, with '10+0' being the most frequent (38%). There are 400 unique time increments recorded.
• white_id: The unique identifier for the player controlling the white pieces. There are 9,438 unique white player IDs.
• white_rating: The Elo rating of the white player at the start of the game. Ratings range from 784 to 2700, with an average of 1,600.
• black_id: The unique identifier for the player controlling the black pieces. There are 9,331 unique black player IDs.
• black_rating: The Elo rating of the black player at the start of the game. Ratings range from 789 to 2723, with an average of 1,590.
• moves: The full sequence of chess moves played within the game. There are 18,920 unique move sequences recorded, with 'e4 e5' being the most common start.
• opening_code: The standard chess opening code (e.g., A00, C00). 'A00' is the most common code, representing 5% of games. There are 365 unique opening codes.
• opening_moves: The number of moves defining the opening sequence. The average is 4.82 moves, ranging from 1 to 28 moves.
• opening_fullname: The full name of the chess opening, such as 'Van't Kruijs Opening' or 'Sicilian Defense'. There are 1,477 unique full names, with 'Van't Kruijs Opening' being the most common (2%).
• opening_shortname: A common short name for the chess opening. 'Sicilian Defense' is the most frequent (13%). There are 128 unique short names.
• opening_response: Details any specific response to the opening. This column has a significant number of missing values (94%), with 'Declined' being a recorded response.
• opening_variation: Specifies the variation within an opening. This column is missing for 28% of games, with '#2' being the most common variation (4%) among recorded entries.

Distribution

Usage

• Determining the percentage of games won by white, black, or ending in a draw.
• Identifying the most frequently used opening moves based on who won the game.
• Investigating the correlation between player ratings and game outcomes, and whether this varies by piece colour.
• Pinpointing the user who won the most games and understanding their performance when rated higher than their opponent.
• Analysing popular opening strategies and their success rates.

Coverage

License

Who Can Use It

• Chess Enthusiasts and Analysts: To delve into game statistics, study opening theory, and understand player performance trends.
• Data Scientists and Machine Learning Engineers: For building predictive models for game outcomes, rating changes, or identifying optimal strategies.
• Researchers: For academic studies on game theory, player behaviour, and the impact of various factors in chess.
• Game Developers: To inform AI development for chess games or to design features based on real-world player data.

Dataset Name Suggestions

• Lichess Chess Game Analytics
• Online Chess Game Records
• Lichess Player Performance Dataset
• Chess Opening Statistics from Lichess
• Lichess Game Data Collection

Attributes