Chess Multiverse Opening Intelligence Database (20M Lichess subset)
收藏Zenodo2026-03-29 更新2026-05-26 收录
下载链接:
https://zenodo.org/doi/10.5281/zenodo.19307100
下载链接
链接失效反馈官方服务:
资源简介:
🚀 Live Interactive Tool
Explore the dataset visually without writing any code. The full 20M+ game database powers the Opening Analytics Module on the Chess Multiverse platform: 👉 Launch the Opening Intelligence Platform
📊 Dataset Overview
The Chess Multiverse Opening Intelligence Database is a high-fidelity, open-source dataset containing statistical performance metrics for 15,013 distinct chess opening lines.
The data was extracted, filtered, and processed from a subset of 20 million standard-rated games played on Lichess during the 2024 calendar year. Processed by Chess Multiverse Lab v1.1 (Stable), this dataset strips out engine evaluation noise to focus exclusively on human-vs-human practical outcomes across various ELO brackets.
📁 File Specifications
Filename: openings_2024_12_depth5_20M.json
Format: JSON (Array of Objects)
Record Count: 15,013
Data Vintage: 2024
🗄️ Data Schema (Data Dictionary)
Each object in the JSON array represents a specific opening sequence and contains the following key-value pairs:
Key
Data Type
Description
Example
eco
String
The standard Encyclopedia of Chess Openings (ECO) code.
"C44"
opening
String
The standardized, formal name and variation of the opening.
"Scotch Game: Lolli Variation"
moves
String
The sequence of moves leading to the position, formatted in Universal Chess Interface (UCI) notation.
"e2e4 e7e5 g1f3 b8c6 d2d4 e5d4..."
games
Integer
The total number of games in the 20M subset that reached this exact position.
22326
white_win_rate
Float
The percentage of games won by White (expressed from 0.0 to 1.0).
0.527
black_win_rate
Float
The percentage of games won by Black (expressed from 0.0 to 1.0).
0.422
draw_rate
Float
The percentage of games that ended in a draw (expressed from 0.0 to 1.0).
0.051
avg_rating
Integer
The average ELO rating of the players who reached this position.
1507
📄 Sample Data
Below is a brief sample of the JSON structure demonstrating how the data is natively formatted:
[
{
"eco": "C44",
"opening": "Scotch Game: Lolli Variation",
"moves": "e2e4 e7e5 g1f3 b8c6 d2d4 e5d4 f3d4 c6d4 d1d4 d7d6",
"games": 22326,
"white_win_rate": 0.527,
"black_win_rate": 0.422,
"draw_rate": 0.051,
"avg_rating": 1507
},
{
"eco": "B32",
"opening": "Sicilian Defense: Löwenthal Variation",
"moves": "e2e4 c7c5 g1f3 b8c6 d2d4 c5d4 f3d4 e7e5 d4c6 b7c6",
"games": 17417,
"white_win_rate": 0.476,
"black_win_rate": 0.482,
"draw_rate": 0.042,
"avg_rating": 1746
}
]
🧪 Methodology & Processing
To ensure the highest level of statistical integrity for academic and analytical use:
Game Selection: Only standard-rated, human-vs-human Lichess games were included. Bullet, hyper-bullet, and variant games were excluded to maintain opening phase validity.
Depth Parsing: Opening lines were mapped up to an average depth of 5 full moves (10 ply), capturing key middle-game transitions.
Statistical Aggregation: Win/loss/draw rates are strictly empirical, reflecting the practical edge in human play rather than perfect engine evaluations.
💻 Usage & Implementation
This JSON dataset is lightweight and structured for immediate parsing in standard data science pipelines.
Python Example (Pandas):
import pandas as pd
import json
# Load the dataset
with open('openings_2024_12_depth5_20M.json', 'r') as file:
data = json.load(file)
# Convert to DataFrame
df = pd.DataFrame(data)
# Find the highest winning openings for Black with at least 5,000 games played
solid_black_lines = df[(df['games'] >= 5000)].sort_values(by='black_win_rate', ascending=False)
print(solid_black_lines.head())
🤝 Contributing
Contributions to improve the dataset, refine parsing algorithms, or build new UI modules are welcome!
Fork the repository.
Create your feature branch (git checkout -b feature/DataRefinement).
Commit your changes (git commit -m 'Add specific refinement').
Push to the branch (git push origin feature/DataRefinement).
Open a Pull Request.
If you spot a statistical anomaly or an incorrect ECO classification, please open an Issue with the relevant UCI string.
📝 Citation & Academic Use
If you use this dataset in data science projects, chess engine development, or statistical research, please cite it using the following DOI:
APA Format:
Varshney, S. (2026). Chess Multiverse Opening Intelligence Database (20M Lichess subset) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.19307100
BibTeX:
@dataset{varshney_2026_chess_db,
author = {Varshney, Sparsh},
title = {Chess Multiverse Opening Intelligence Database (20M Lichess subset)},
month = {March},
year = {2026},
publisher = {Zenodo},
doi = {10.5281/zenodo.19307100},
url = {[https://doi.org/10.5281/zenodo.19307100](https://doi.org/10.5281/zenodo.19307100)}
}
👨🔬 Lead Researcher & Principal Developer
Sparsh Varshney • Founder • Data Scientist • Researcher
Sparsh is a medical researcher and open-source data scientist currently pursuing a Bachelor of Ayurvedic Medicine and Surgery (BAMS) at Uttarakhand Ayurved University. By bridging rigorous academic research methodologies with modern web development and AI, Sparsh builds high-fidelity datasets and analytical tools. He is the founder of Chess Multiverse and Amidha Ayurveda, developing specialized data platforms across multiple disciplines.
About the Creator
LinkedIn
ORCID
GitHub
⚖️ License
This dataset is published under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license. You are free to share and adapt the material for any purpose, even commercially, provided you give appropriate credit and distribute your contributions under the same license.
提供机构:
Zenodo创建时间:
2026-03-29



