How to backtest on Hyperliquid

To backtest on Hyperliquid, get history from a source that reaches past the API's limit of 5000 candles per interval, then simulate with Hyperliquid's real fees and hourly funding. Free Parquet candles cover every perp and spot pair hourly from 25 May 2025, and Ironflow's API serves 1 minute bars and every trade from the same date.

The answer#

• Hourly strategy: use the free hourly candles. No key, no account.
• Intraday strategy: use /v1/candles at 1 minute, or build bars from every trade with SQL.
• Always: charge fees on every trade and funding on every position held across the hour.
• Then: run it on a paper account with live prices before real money.

What each source covers#

API reach depends on the plan: 24 hours without a key, 14 days with a free key, 90 days on Explorer and 1 year on Builder. The free files have no such limit. Hyperliquid's info endpoint returns only the most recent 5000 candles per interval.

The exact calls#

Hourly candles in pandas#

One month of the free dataset, then a moving average cross with the base taker fee. We ran this on the October 2026 file: 744 hourly BTC bars.

import pandas as pd

URL = "https://huggingface.co/datasets/ironflow/hyperliquid-candles-1h/resolve/main/data/2026-10.parquet"
df = pd.read_parquet(URL, columns=["timestamp", "coin", "open", "high", "low", "close", "volume"])
btc = df[df.coin == "BTC"].sort_values("timestamp").set_index("timestamp")

TAKER_FEE = 0.00045  # Hyperliquid's base perp taker fee, 0.045%
c = btc["close"]
fast, slow = c.rolling(24).mean(), c.rolling(96).mean()
position = (fast > slow).astype(int).shift(1).fillna(0)
returns = position * c.pct_change().fillna(0) - position.diff().abs().fillna(0) * TAKER_FEE
print((1 + returns).prod() - 1)

Files are named by month, data/YYYY-MM.parquet. Load each month you need and concatenate. Needs pandas and pyarrow.

1 minute candles over REST#

curl 'https://api.ironflow.sh/v1/candles?market=BTC-PERP&interval=1m&from=-6h'

Build your own bars from every trade#

One SQL query over hl.trades gives any bar size or filter, with taker volume. Keyless reads the last 24 hours.

curl -G 'https://api.ironflow.sh/v1/query' --data-urlencode "sql=SELECT toStartOfHour(ts) AS hour, argMin(price, (ts, tid)) AS open_px, max(price) AS high_px, min(price) AS low_px, argMax(price, (ts, tid)) AS close_px, round(sum(notional)) AS volume_usd FROM hl.trades WHERE market = 'BTC-PERP' AND ts >= now() - INTERVAL 6 HOUR GROUP BY hour ORDER BY hour"

Fees and funding#

A backtest that skips these is optimistic. From Hyperliquid's fee schedule on 9 November 2026:

• Perps at the base tier: 0.045% taker, 0.015% maker.
• Spot at the base tier: 0.070% taker, 0.040% maker.
• Tiers are set by rolling 14 day volume, and HIP-3 markets add a deployer fee that varies by market.
• Funding settles every hour. The free funding dataset has every settled rate.

What candles cannot show#

• A bar has no order book, so it cannot tell you slippage. Fill your simulation at the next open with a conservative buffer.
• Liquidations and stops can fill far from the close of a bar. Test with 1 minute data or ticks for anything that uses tight stops.
• A strategy that needs order book depth needs book data. See the order book guide.

Test it live on paper#

A paper box is your own simulated Hyperliquid account with 10,000 USD and live prices. It speaks the Hyperliquid API, so the same bot runs against it through the official SDK's base_url.

curl -X POST https://api.ironflow.sh/v1/ironbox

No key is needed for a 24 hour trial. The paper trading guide shows the setup.

Questions#

Related guides#

• Hyperliquid historical data: API, S3 archive, SQL and free downloads
• Hyperliquid candles and OHLCV data
• Hyperliquid funding rates
• Hyperliquid paper trading
• Free Hyperliquid datasets

References#

• Hyperliquid hourly candles dataset. Hourly OHLCV for every perp and spot pair, CC BY 4.0.
• Hyperliquid funding rates dataset. Every settled funding rate, CC BY 4.0.
• Hyperliquid docs: Fees. The fee tiers used above.
• Hyperliquid docs: Info endpoint. The candleSnapshot limit.
• Ironflow docs: Rate limits. History and requests by plan.