Backtest your Hyperliquid strategy
Keel is a backtesting tool for Hyperliquid perpetuals. It runs your strategy over Hyperliquid's own price and funding history and charges trading fees, slippage and hourly funding on every position, so you see what the rules would have made after costs. You build the strategy in plain words or blocks with no code, and the same strategy can then paper trade or run live from your own Hyperliquid account.
For crypto backtesting in general, from the data a test needs to each step in order, read the guide. It also lets you try library strategies first.
Funding Carry (Cross-Sectional, Hyperliquid)
After fees, slippage and funding, on the default settings.
What a Hyperliquid backtest shows you
One strategy from the Keel library: the rules it follows, and how much of its result came from price and how much from funding.
The rules, as blocks
Funding Carry (Cross-Sectional, Hyperliquid) ranks the most traded Hyperliquid perps by their funding rate once a day. It shorts the coins where longs pay the most to hold and buys the coins where shorts pay, so the book is balanced long and short. Read the blocks from the top: price and funding data come in, the funding rate is flipped and smoothed, ranked across coins, then turned into position sizes. Switch to Code to see the same strategy as text.
Configuration
Globals(target_timeframe="1d")
Universe(mode="top_volume", top_n=100, market="perp", lookback="90d", exclusions=["PURR"], resolved=["AAVE", "ADA", "AERO", "ALGO", "APE", "APT", "AR", "ARB", "ASTER", "AVAX", "AXS", "BCH", "BIO", "BLUR", "BNB", "BTC", "CAKE", "CC", "CHIP", "CRV", "DASH", "DOGE", "DOT", "DYDX", "EIGEN", "ENA", "ETC", "ETH", "ETHFI", "FARTCOIN", "FET", "FIL", "GRAM", "GRASS", "HBAR", "HYPE", "ICP", "INIT", "INJ", "IP", "JTO", "JUP", "KAITO", "LDO", "LINK", "LIT", "LTC", "MEGA", "MEME", "MET", "MON", "MORPHO", "NEAR", "NIL", "ONDO", "OP", "ORDI", "PAXG", "PENDLE", "PENGU", "PNUT", "POL", "POPCAT", "PUMP", "PYTH", "RENDER", "RUNE", "SEI", "SKY", "SOL", "SPX", "STABLE", "STRK", "SUI", "SYRUP", "TAO", "TIA", "TON", "TRUMP", "TRX", "UNI", "VIRTUAL", "VVV", "W", "WIF", "WLD", "WLFI", "XLM", "XMR", "XPL", "XRP", "ZEC", "ZEN", "ZORA", "ZRO", "kBONK", "kNEIRO", "kPEPE", "kSHIB"], resolved_at="2026-07-07T19:46:36.522366+00:00")
Execution(rebalance="buffered", buffer_threshold=0.15, buffer_mode="relative", rebalance_method="to_edge", min_trade_size=0)
Pipeline([
PriceDataLoader(),
FundingDataLoader(),
NegateTransform(),
EWMATransform(window=10),
CrossSectionalZScore(),
ForecastScaler(avg_abs_target=10),
ForecastCapper(limit=20),
ForecastWeightNormalizer(target_leverage=1),
], name="funding_carry")
Where the return came from
Keel books funding into the equity curve as it is paid, and also runs the same backtest without it, so you can see what funding did to the result. Here is that split for this strategy and for a momentum strategy from the same library.
Funding is most of the carry strategy's result. Run without it, the same rules would barely be worth trading. The momentum strategy is the other way round: it tends to hold coins that are rising, where longs pay to hold, so funding took a real bite out of a good result. A backtest that skips funding gets both of these wrong. Hyperliquid, August 2024 to September 2026, default settings, after trading fees and slippage in every column.
See the full backtest, monthly returns and every setting testedHow to backtest a Hyperliquid strategy
From an idea to a result you can trust, in the Keel app. No code, no data downloads.
- 01
Start from an idea or a library strategy
Describe the strategy in plain words and Keel's AI builds it from blocks you can read and change. Or open a strategy from the library and adjust it. You never write code unless you want to.
- 02
Pick the coins and the bar size
Test one coin or a whole basket: a list you choose, or the top Hyperliquid perps by trading volume. Bars run from 5 minutes to 1 day.
- 03
Run it over the dates you choose
Keel replays the strategy bar by bar on Hyperliquid history. Every fill pays a 4.5 bps fee and 4.5 bps of slippage, and every open position pays or earns the funding Hyperliquid actually charged.
- 04
Read what happened
You get the equity curve, the drawdowns, every trade, and your return split into what came from price and what came from funding. Ask the AI why a month went badly.
- 05
Change one thing and run it again
Every run is kept, so you can compare versions side by side. To check a result holds up, run it again on dates you held back.
- 06
Paper trade it, then go live
Paper trading runs the saved strategy forward on live prices with no money at risk. When you're ready, connect your Hyperliquid account and the same strategy runs live. Your money stays in your account: Keel can place trades there but can never withdraw.
The Hyperliquid data behind every backtest
Keel collects its own Hyperliquid history and keeps it current, so there is nothing to download or store.
Price bars from every trade
Keel keeps its own record of every trade on Hyperliquid and builds bars from it: 5, 15 and 30 minutes, 1 to 12 hours, and daily. Your strategy picks one bar size and runs on it.
Hourly funding
The funding rate Hyperliquid settled for each coin, every hour. A backtest charges it on every open position at the rate that actually applied, so a long pays when funding is positive and a short collects it.
Open interest and order flow
Open interest, premium and predicted funding per coin, plus order flow built from trades: taker buying and selling, large prints, VWAP and dollar volume. Each is a data block you can use in a strategy. Order flow starts in March 2025.
How far back it goes
BTC, ETH and SOL go back to mid-2024. Other coins start when they listed on Hyperliquid, so a new listing has a short history. Bars of 15 minutes and up reach the full history; 5-minute bars start in March 2025.
Every strategy in the Keel library is tested over the same window, August 2024 to September 2026, so their results line up. Your own backtests can use any window the data covers.
Fees, funding and slippage, charged as you go
Most backtests that look great and then lose money live skipped one of these. Keel charges all three on every run.
Trading fees
Every fill pays 4.5 bps (0.045%), Hyperliquid's base taker fee. If you trade at a lower fee tier, change it in the backtest settings.
Slippage
Every fill is also priced 4.5 bps against you: buys a little above the bar price, sells a little below. If the strategy's edge is thin, raise it and see whether the result survives.
Funding
Booked into the equity curve as it is paid, at Hyperliquid's real rate for each coin. A strategy that holds crowded longs pays for it, and a carry strategy collects it. The result shows how much of your return each side made.
The same limits as live
Positions are sized across all your coins together, and gross leverage is capped at 10x in a backtest and live alike. A test can't lean on leverage the live strategy wouldn't be allowed.
The backtest and the live strategy share one definition, so what you tested is what runs. For the details of how a strategy is checked, compiled and run, read how Keel works.
After the backtest, paper trade it on live prices
A backtest replays the past. The next check is the present: Hyperliquid paper trading runs the same saved strategy forward on live Hyperliquid prices, with simulated fills and no money at risk. Think of it as a Hyperliquid simulator that uses mainnet prices instead of the testnet. When the paper record looks right, switch it on in your own account and it becomes your Hyperliquid bot.
How paper trading works- Fills are priced from live Hyperliquid prices, with fees, slippage and real hourly funding.
- No wallet, exchange account or API key needed. A Keel account and a saved strategy are enough.
- The run's page shows how closely it is tracking its backtest over the same dates.
- The testnet is a separate exchange with its own prices and funding. Paper trading uses the real ones.
- When you're ready, the same strategy goes live from your own account. Nothing forces the move.
Backtesting Hyperliquid in Python instead
Plenty of traders backtest Hyperliquid in Python, and for some it is the right call. Here is what each route asks of you.
Doing it yourself in Python
Suits you if you code, want every detail in your own hands, or trade faster than 5-minute bars.
- You start from Hyperliquid's Python SDK or one of the open-source backtest libraries on GitHub.
- History is the first job. Hyperliquid's API serves only about the last 5,000 candles per coin and bar size, and open interest only as of now. Anything longer you collect and store yourself, or buy from a data vendor.
- You write the cost model: fees, slippage and hourly funding on every position. Funding is the part most often left out, and on Hyperliquid it can decide the result.
- Going live usually means a second program, and keeping it in step with the one you tested.
Backtesting in Keel
Suits you if you want to test trading ideas, not build and run the data and cost plumbing.
- Hyperliquid price, funding, open interest and order-flow history is already collected and kept current.
- Fees, slippage and funding are charged on every run, and portfolios across many coins work out of the box.
- The strategy you tested is the one that paper trades and runs live.
- The limits: bars from 5 minutes up, so no tick-level or market-making strategies. Hyperliquid's native crypto perps only for now. Strategies are built from Keel's 262 blocks, not arbitrary code.
If you write Python but don't want to build the data and cost plumbing, the Keel CLI guide shows how to write strategies in your editor and run them on the same engine from a terminal or a notebook.
Calculators to use around a backtest
Quick answers to the questions that come up before and after a run. Free, no sign-up.
Funding calculator
What holding a Hyperliquid perp costs or earns in funding over hours, days or weeks, with the APR.
Open the calculatorOverfit check
A score for how likely a backtest result is to be overfit, from its returns. Runs in your browser.
Open the calculatorLiquidation calculator
The liquidation price of a long or short from entry and leverage, using the coin's real maintenance margin.
Open the calculatorThe numbers
| cost multiple (Cost multiple) | Sharpe | CAGR | Max drawdown | Total return | days |
|---|---|---|---|---|---|
| 0 | 1.36 | 68.0% | −37.0% | 201.0% | 775 |
| 1 | 1.17 | 54.3% | −38.8% | 151.0% | 775 |
| 2 | 0.61 | 19.3% | −43.7% | 45.4% | 775 |
| 3 | −0.32 | −22.4% | −69.1% | −41.7% | 775 |
| 4 | −1.61 | −57.8% | −89.0% | −84.0% | 775 |
| 6 | −5.07 | −92.8% | −99.7% | −99.6% | 775 |
Keel's study of Momentum Strategy on Hyperliquid · computed Nov 16, 2026 · data Aug 15, 2024 to Sep 29, 2026
What data does a Hyperliquid backtest on Keel use?
Hyperliquid's own data, collected by Keel: price bars built from every trade on the venue, hourly funding rates, open interest, premium, and order flow such as taker buying and selling. Strategies can run on bars from 5 minutes to 1 day. Nothing needs downloading or cleaning. Stock and index perps are HIP-3 markets, and Keel doesn't support them yet. Support is coming soon.
How far back does the Hyperliquid historical data go?
BTC, ETH and SOL go back to mid-2024. Every other coin starts when it listed on Hyperliquid, so recent listings have less history. Bars of 15 minutes and up reach the full history, while 5-minute bars and order flow start in March 2025. When you run a backtest, Keel works out how far back your coins' data goes.
How are fees, funding and slippage handled?
Every fill pays a 4.5 bps fee, Hyperliquid's base taker rate, and 4.5 bps of slippage against you. You can change both in the backtest settings. Funding is charged or paid on every open position at the rate Hyperliquid actually settled, and the result splits your return into price and funding.
Can I backtest a portfolio of Hyperliquid coins?
Yes. A strategy can trade a list of coins you choose or the top Hyperliquid perps by volume, and Keel sizes positions across all of them together, so correlation, turnover and costs show up the way they would in a real account.
How close will live trading be to the backtest?
The live strategy runs the same definition you backtested, with the same sizing rules and leverage cap, so there is no rewrite between the two. Live results still differ: real fills, market impact and new market conditions don't match any model exactly. Paper trading first shows you the gap on live prices. The How Keel works page explains what stays identical between the two.
Is there a Hyperliquid simulator or paper trading mode?
Yes. Paper trading runs a saved strategy forward on live Hyperliquid prices with simulated fills, fees and real hourly funding, and no wallet. It shows how the run tracks its backtest, and you can move it live whenever you choose.
Can I backtest Hyperliquid in Python instead?
Yes, and if you want full control or tick-level detail it is a reasonable choice. Expect to collect and store the history yourself, since Hyperliquid's API serves only recent candles, and to write the fee, slippage and funding model. If you like working in code but not building that, Keel's command-line tool (pipx install keel-trade) runs backtests on the same engine and data from a terminal or notebook.
What does it cost?
Backtesting is free to start. The free plan includes 25 backtests every 30 days, plus 50 in your first week, and one paper strategy on a timeframe of 1 hour or slower. Paid plans start at $49 a month, with more backtests and lower fees on live trades. The pricing page has the full limits.
Run your first Hyperliquid backtest
Starting is free. Describe a strategy or open one from the library, and see what it would have made after costs.