Dataset · phobia76
Polymarket crypto platform order book backup
phobia76/pmxt-l2-dump
An unofficial mirror of hourly Polymarket order book dumps in parquet format.
What's inside
The repository is a backup of public PMXT archives containing hourly order book snapshots from Polymarket. The dataset size ranges from 100 billion to 1 trillion entries stored as parquet files. The description is provided in English and Chinese, and the data is distributed under the CC BY license.
What you can do with it
- Training tabular and time-series models for predicting liquidity and price movement in prediction markets
- Backtesting algorithmic trading strategies and analyzing market microstructure on historical data
- Tasks
- Tabular classification
- Languages
- English, Chinese
- Size
- 100B–1T rows
- Downloads this month
- 21K
- Likes
- 7
How to load it
from datasets import load_dataset
ds = load_dataset("phobia76/pmxt-l2-dump")