Hyperliquid datasets: free trades, liquidations, funding and candles - Ironflow

For bulk history you do not need the API. The same data is published as free Parquet datasets under CC BY 4.0, on Hugging Face and most of them on Kaggle. Credit “Ironflow (ironflow.sh)” when you publish work built on them.

Dataset	What it holds	From
Hyperliquid Trades	Every perp and spot trade, one row per trade with the buyer, the seller and the taker side. A new file every day.	3 October 2026
Hyperliquid Liquidations	Every liquidation on native and HIP-3 perps, as events and as fills, from the `liquidation` field Hyperliquid puts on each fill	25 May 2025
Hyperliquid Funding Rates	Every settled hourly funding rate, native and HIP-3 perps	May 2023
Hyperliquid Hourly Candles	Hourly OHLCV for every perp and spot pair, built from fills, delisted markets included	25 May 2025
Hyperliquid Open Interest, Mark and Oracle Prices	Hourly open interest, mark and oracle price for every perp	May 2023
Hyperliquid HIP-4 Outcome Markets	Every HIP-4 outcome fill and every contract with its event, outcome and settlement. A new file every day.	1 August 2026

Each dataset page describes its columns, coverage and checks.

Load one in Python

import pandas as pd

# One row per liquidation event (needs huggingface_hub installed)
events = pd.read_parquet("hf://datasets/ironflow/hyperliquid-liquidations/data/events/")
print(events.groupby("market")["notional_usd"].sum().sort_values(ascending=False).head(10))

Column names differ per dataset: check the dataset page before you join two of them.

Datasets or the API

Datasets: the whole history in one download, for research and backtests. Updated daily (trades, outcomes) or with each release (the others).
API: the last seconds to a year, filtered by market, wallet and time, plus live streams, the liquidation map and SQL. Start with the Quickstart.