Free crypto backtesting software that works with your AI
Keel is free crypto backtesting software for Hyperliquid perpetuals. You can run backtests from Claude, ChatGPT, Cursor or Claude Code by connecting Keel's hosted MCP server, or in the Keel app with its AI builder or blocks, and every run charges fees, slippage and hourly funding. The free plan includes 50 backtests a week with full reports.
- 1Customize → Connectors → Add custom connector
- 2Paste https://mcp.usekeel.io/mcp and choose Connect
- 3Sign in or sign up on the Keel page that opens (no account needed beforehand)
https://mcp.usekeel.io/mcpSign in in the browser when asked. No API keys. Connect → sign in to Keel → Allow
Two ways to tell, and either one settles it. A connection that is already live is not improved by signing in again.
In Claude Keel is listed where you added it (Customize → Connectors → Add custom connector) once the sign-in has finished.
Ask the agent keel_account_status answers with the Keel account the session is signed in as, which nothing but a live connection can return. If the agent says it has no Keel tools, the sign-in did not finish.
Use keel_account_status and tell me which Keel account this session is signed in as.Not yet verified against a Claude version. These steps follow Anthropic's own documentation, read on 2026-09-16.
Two ways to backtest crypto for free
Both routes run the same backtest engine on the same Hyperliquid data, and both are covered by the free plan. Pick the one that matches where you already work.
Backtest from Claude, ChatGPT, Cursor or Claude Code
Add Keel's hosted MCP server to your assistant once, sign in to Keel in your browser, and the assistant gets 29 tools: search 251 strategy blocks, build a strategy, run a backtest and read the result back to you.
Then ask in plain words. Something like: backtest a MACD crossover on the top 15 Hyperliquid perps on 4-hour bars, long only when price is above its 200-bar average. The assistant builds it, runs it on Keel and tells you what happened after costs. You can keep going in the same chat: change the bar size, add a filter, compare two versions.
There is nothing to install and no API key. The agent builds, tests and reads. When you want to run a strategy on your account, it hands it to the Keel app and you take it from there.
Set-up guides, one per assistant
Backtest in the app with the AI builder or blocks
Sign up free and describe the strategy to Keel's own AI builder, or open a strategy from the library and change it. Every strategy is a set of blocks you can read: data in, signals, filters, sizing. You can edit the blocks directly if you would rather not chat.
Pick the coins and the dates, press run, and you get the equity curve, the drawdowns, every trade and how much of the return came from price and how much from funding. Every run is kept, so you can compare versions side by side.
From the same screen, a strategy you like can paper trade on live prices, and later run live from your own Hyperliquid account. Keel can place trades there but can never withdraw.
What free includes
The same numbers the pricing page shows. The free plan has no time limit.
Backtests a week
From the app or from your assistant. Both draw on the same allowance.
Backtest compute a week
Run time on the backtest workers. A basket of many coins on short bars uses more of it than one coin on daily bars.
AI builder messages a week
For the AI builder inside the Keel app. Your own assistant uses its own plan.
Paper strategies
Run a tested strategy forward on live prices with no money at risk.
Live strategies
Run from your own Hyperliquid account, at a 5 bps fee on live trades.
Backtest reports
Equity curve, drawdowns, every trade, and the price and funding split.
When you outgrow it, paid plans start at $29 a month with more backtests, more paper and live strategies, and lower fees on live trades. See pricing for every limit.
One free backtest, start to finish
Here is the kind of strategy most people test first: an indicator rule. This one is published in the Keel library, so you can open it, change it and run it yourself on the free plan.
It holds a coin long while its MACD histogram is above zero and its price is above its 200-bar average, and sits in cash otherwise. It trades a basket of 15 of the most traded Hyperliquid perps on 4-hour bars, with equal weights. Read the blocks from the top, or switch to Code to see the same strategy as text.
Configuration
Factories
Globals(target_timeframe="4h")
Universe(mode="top_volume", top_n=15, market="perp", resolved=["BTC", "DOGE", "ETH", "GRAM", "HYPE", "LINK", "LIT", "PUMP", "SOL", "SUI", "VVV", "XPL", "XRP", "ZEC", "kPEPE"], resolved_at="2026-07-06T20:36:37.412320+00:00")
Execution(rebalance="on_change")
trend_vs_200 = Pipeline([
{
"px": [EWMA(window=2)],
"trend": [EWMA(window=200)],
},
Crossover(fast_key="px", slow_key="trend"),
], name="trend_vs_200")
Pipeline([
PriceDataLoader(),
Store("ohlcv"),
{
"entries": [
Pipeline([
{
"const": [ConstantForecast(value=1.0)],
"setup": [
Pipeline([
{
"macd_up": [MACD(fast_period=12, slow_period=26, signal_period=9), AboveThresholdFilter(threshold=0.0)],
"uptrend": [trend_vs_200, AboveThresholdFilter(threshold=0.0)],
},
MaskAnd(),
], name="setup"),
],
},
ApplyMask(score_signal="const", filter_signal="setup"),
Store("macd_entry"),
], name="entries"),
],
"exits": [
Pipeline([
{
"const": [ConstantForecast(value=1.0)],
"cross_down": [MACD(fast_period=12, slow_period=26, signal_period=9), BelowThresholdFilter(threshold=0.0)],
},
ApplyMask(score_signal="const", filter_signal="cross_down"),
Store("macd_exit"),
], name="exits"),
],
},
TradeManager(entries="macd_entry", prices="ohlcv"),
{
"macd_cross_down": [Load("macd_exit"), Exit()],
},
Exposure(),
EqualWeightSizer(target_leverage=1.0, max_weight=0.2),
], name="macd_crossover_crypto")
On the default settings it made a Sharpe of 1.87 and +408.8% total return, with a worst drawdown of 33.3%, after fees, slippage and funding. The same rules on daily bars made a Sharpe of 0.48 and +25.8% total return. That is the point of running the test before trading: here the bar size mattered more than the MACD settings.
Open the strategy, its monthly returns and every setting testedFees, slippage and funding on every run
Free backtesting tools often skip costs, and on crypto perps the costs can be the whole story. A strategy that trades a lot loses its edge to fees. A strategy that holds crowded longs pays funding every hour. Keel charges all of it by default, on the free plan and the paid ones alike.
Trading fees
Every fill pays 4.5 bps, 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. Raise it to see whether a thin edge survives worse fills.
Hourly funding
Every open position pays or earns the funding rate Hyperliquid actually settled for that coin and hour, booked into the equity curve as it happens.
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.
What Keel's free backtesting does not do
Keel is built for one kind of trader: someone testing systematic strategies on Hyperliquid perpetuals. If that is not you, another tool will serve you better, and the comparison page lists the main kinds.
Hyperliquid perps only
Keel tests Hyperliquid's native crypto perpetuals. There is no spot, no forex and no stock market data. Stock and index perps are HIP-3 markets, and Keel doesn't support them yet. Support is coming soon.
Bars, not ticks
Strategies run on bars from 5 minutes to 1 day. Tick-level, high-frequency and market-making strategies are out of scope.
Blocks, not arbitrary code
Strategies are built from Keel's 251 blocks, which keeps every strategy readable and checkable. If you need a custom model the blocks can't express, a code-first library is the better fit.
History has a start date
BTC, ETH and SOL go back to mid-2024. Other coins start when they listed on Hyperliquid, and 5-minute bars start in March 2025.
No manual chart replay
Keel tests rules, not discretionary trading. If you want to step through candles and practise your own entries by hand, a chart replay tool is what you are after.
A backtest is not a forecast
Costs are charged, but real fills, market impact and new conditions never match a model exactly. Paper trading on live prices is the check between the backtest and real money.
Other free backtesting software, and who it suits
Free backtesting software comes in a few kinds. Open-source Python libraries are free and flexible if you write code and source your own data. Charting platforms have testers built into the chart. Bot platforms let you test a preset bot before you switch it on. Chart replay tools let you practise by hand.
Keel's place is the systematic trader on Hyperliquid perps who would rather describe a strategy than code its plumbing. If you are weighing the options, the comparison sets out the main tools in each kind, with what each one is for. If you are new to backtesting, the guide covers how to do it properly, whatever tool you pick.
Questions about free crypto backtesting
Can ChatGPT backtest a trading strategy?
Not well on its own. ChatGPT can explain a strategy or write backtest code, but it has no market history or cost model to run it against, so its answers about results are guesses. Connected to Keel's MCP server, it builds the strategy and runs a real backtest on Hyperliquid data with fees, slippage and funding, then reads you the result.
Can Claude backtest a crypto strategy?
Yes, once you connect Keel. In Claude or Claude Code you add Keel's hosted MCP server, sign in in the browser, and ask for a backtest in plain words. Claude builds the strategy from Keel's blocks, runs it on Hyperliquid history and explains the result.
Is Keel's backtesting really free?
Yes. The free plan includes 50 backtests a week with full reports and one paper strategy. Paid plans start at $29 a month if you need more backtests or lower fees on live trades.
Do I need to code to backtest crypto strategies?
No. You can describe the strategy to your assistant or to Keel's AI builder, or assemble the blocks yourself. If you prefer code, the keel-trade command-line tool runs backtests on the same engine from a terminal.
Which coins and data can I backtest?
Hyperliquid's native crypto perpetuals, on price bars built from every trade, with hourly funding, open interest and order flow available as data blocks. A strategy can trade one coin or a basket, such as the most traded perps by volume.
How realistic are the backtest results?
Each run charges fees and slippage on every fill and the real hourly funding on every position, with the same leverage cap as live. The strategy you backtest is the same definition that paper trades and runs live, so nothing is rewritten in between. Results still differ live, which is why paper trading comes first.
What is the best free backtesting tool for crypto?
It depends on how you work. Python coders often pick an open-source library, chart traders use their charting platform's tester, and people who want an AI to do the building can use Keel. Our comparison of crypto backtesting software sets out who each kind suits.
Run your first free backtest
Connect your assistant and ask for a backtest, or sign up and build one in the app. Either way it is free to start.