Chess Games from High-Level Players, 14 Million Games with 1.2 Billion Moves

The Chess Games dataset by angeluriot is a collection of 14 million high-level chess games for training machine learning models. It contains 1.2 billion moves played between 1600 and 2024, with a mean player ELO of 2388 and a mean of 84 moves per game. The dataset was last updated on Hugging Face in November 2024.

Use Cases

• Train chess move prediction models based on the sequence of 1.2 billion moves.
• Analyze game outcomes and resignation patterns based on the high-level player data.
• Develop models to estimate player strength based on the mean ELO of 2388.
• Study game length and complexity based on the mean of 84 moves per game.

Strengths

• Large scale with 14 million games and 1.2 billion moves.
• High-quality source data from players with a mean ELO of 2388.
• Long temporal coverage from 1600 to 2024.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The description notes most games are recent, which may indicate a temporal bias toward modern play.

Provenance

Source

Hugging Face user angeluriot.

Collection Method

Collection of high-level chess games, likely aggregated from online platforms.

Time Range

1600 to 2024.

Freshness

Last updated 2024-11-21 00:35:09.

Geography

null