Historical DEX trading data

The following datasets are available for historical DEX trading data. Sign up for a free API key to download the data.

Read the documentation how to get started with Trading Strategy Python library for algorithmic trading .

Available datasets

Name Tag Entries Size Format Last updated Links
Decentralised exchanges exchange_universe 7.5K 4.5 MB JSON
Trading pairs pair_universe 282.8K 35.1 MB Parquet
OHLCV candles, 1 minute candles_1m 693.5M 51.5 GB Parquet
OHLCV candles, 5 minutes candles_5m 410.3M 30.9 GB Parquet
OHLCV candles, 15 minutes candles_15m 264.5M 20.2 GB Parquet
OHLCV candles, 1 hour candles_1h 136.3M 10.6 GB Parquet
OHLCV candles, 4 hours candles_4h 62.6M 4.9 GB Parquet
OHLCV candles, daily candles_1d 19.9M 1.6 GB Parquet
OHLCV candles, weekly candles_7d 4.9M 423.4 MB Parquet
OHLCV candles, montly candles_30d 1.8M 156.0 MB Parquet
XY Liquidity, 1 minute liquidity_1m 680.2M 33.9 GB Parquet
XY Liquidity, 5 minutes liquidity_5m 403.6M 20.2 GB Parquet
XY Liquidity, 15 minutes liquidity_15m 263.0M 13.3 GB Parquet
XY Liquidity, 1 hour liquidity_1h 138.8M 7.1 GB Parquet
XY Liquidity, 4 hours liquidity_4h 65.8M 3.4 GB Parquet
XY Liquidity / TVL, daily liquidity_1d 26.5M 1.2 GB Parquet
XY Liquidity / TVL, weekly liquidity_7d 6.2M 339.3 MB Parquet
XY Liquidity / TVL, monthly liquidity_30d 2.1M 128.8 MB Parquet
Top momentum, daily top_movers_24h 500.0 660.3 KB JSON
AAVE v3 supply and borrow rates aave_v3 14.5M 1.2 GB Parquet
Uniswap trades uniswap_trades 2.0B 98.0 GB Parquet

Data logistics

Datasets are distributed in Parquet file format designed for data research. Parquet is a columnar data format for high performance in-memory datasets from Apache Arrow project.

Datasets are large. Datasets are compressed using Parquet built-in Snappy compression and may be considerably larger when expanded to RAM. We expect you to download the dataset, cache the resulting file on a local disk and perform your own strategy specific trading pair filtering before using the data. Uncompressed one minute candle data takes several gigabyte of memory.