Python is the default language for trading bots: pandas and NumPy for data, scikit-learn or PyTorch for models, and an AI assistant that writes it fluently. The hard part isn't the strategy code. It's everything around it: clean historical data, a backtester you trust, a server that never sleeps, order handling that survives a dropped connection, and limits that stop a bug from emptying an account.
The tools below take different shares of that work off your hands:
- Hosted platforms run your bot for you.
- Open-source frameworks give you the engine, and you run the server.
- Broker APIs give you market access, and you build the rest.
| # | Platform | Type | Hyperliquid | Hosting | Starts at |
|---|---|---|---|---|---|
| 1 | Antei | Hosted platform | Yes | Managed | Free (no subscription) |
| 2 | QuantConnect | Hosted platform, open-source engine | No | Managed or self-host | Free (research) |
| 3 | Freqtrade | Open-source framework | Yes | Self-host | Free |
| 4 | Hummingbot | Open-source framework | Yes | Self-host | Free |
| 5 | NautilusTrader | Open-source engine | Yes (in development) | Self-host | Free |
| 6 | Jesse | Framework, paid live plugin | Live only | Self-host | Free; live from $899 |
| 7 | VectorBT PRO | Research library | No | n/a | $25/month |
| 8 | Alpaca | Broker API | No | Self-host | Free API |
| 9 | Interactive Brokers | Broker API | No | Self-host | Free API |
| 10 | MetaTrader 5 | Terminal API | No | Self-host (Windows) | Free |
How we ranked them
- Time from idea to a running bot, including hosting.
- Backtesting you can trust: real data for the market you trade, fees and fills included.
- Paper trading before real money.
- Safety: risk limits outside your code, and keys your code never sees.
- Freedom: any Python package you need, and code you own.
1. Antei: best for hosting a Python bot without running a server
Best for: Python developers who want to write the strategy and let a platform handle the build, backtests, hosting, order handling and risk limits, on Hyperliquid.
You upload a main.py (or a folder or zip with requirements.txt). Antei scans it for malware, builds an immutable version in an isolated sandbox, backtests it, and runs it in paper trading or live on Hyperliquid. Node.js works the same way with main.js and package.json.
What stands out
-
Your stack, not a dialect. Use pandas, NumPy or your own models: any PyPI package with a binary wheel installs from
requirements.txt. Antei's SDK is preinstalled. - Real order handling. The SDK places market, limit and post-only orders through Antei's order system and reports each order's status, fills and fees.
- Backtests that can't touch the internet. Replay up to two years of Hyperliquid history in a sandbox with no network access, so a test never reaches a live service.
- Risk limits outside your code. Allowed markets, capital, stop loss, take profit, maximum drawdown, daily loss, position size and leverage, adjustable on a running deployment without a new upload.
- Keys your code never sees. Antei trades through a Hyperliquid API wallet that can't withdraw; its key is encrypted with AES-256-GCM and never reaches strategy code.
- Build it with AI. Claude Code, Codex or claude.ai can write, upload, backtest and deploy the bot through Antei's MCP server.
Pricing: free to use, with no subscription. Backtests and paper trading cost nothing; live orders carry Antei's builder fee on Hyperliquid, which you approve once before your first live trade.
Watch-outs: trades Hyperliquid perpetual futures only. If you need stocks, options or other exchanges, look at the frameworks and broker APIs below.
2. QuantConnect: best for multi-asset quant research
Best for: quants who want deep historical data, many brokers and research notebooks in one cloud platform.
QuantConnect runs Python (or C#) algorithms on its open-source LEAN engine: research in notebooks, backtest in the cloud, then trade live on cloud nodes or your own machine. Its Mia assistant and an official MCP server help you write and debug.
Pricing: free for research and backtests (no live trading); Researcher $84/month; Team $168/month; live nodes cost extra.
Watch-outs: no Hyperliquid (dYdX is the only DEX on its brokerages list), and live trading means a paid plan plus nodes.
3. Freqtrade: best free open-source crypto bot
Best for: crypto traders comfortable running Docker or a Linux server.
Freqtrade is a mature, free, open-source (GPL-3.0) crypto bot with backtesting, hyperparameter optimisation, dry-run paper trading, a web UI and Telegram control. FreqAI adds machine learning.
Pricing: free; you pay for your own server.
Watch-outs: you host and secure it yourself. Hyperliquid works for spot and futures, but without native market orders and with only about 5,000 historic candles, so Hyperliquid backtests are short.
4. Hummingbot: best for market making
Best for: market makers and liquidity providers.
Hummingbot is an open-source (Apache-2.0) framework for market making and high-frequency strategies, with Hyperliquid spot and perpetual connectors and an MCP server for Claude Code, Codex and Gemini CLI.
Pricing: free and self-hosted. Its Condor AI harness adds a 1 bp builder fee on Hyperliquid.
Watch-outs: a steep learning curve, and it's built for market making more than signal-driven strategies.
5. NautilusTrader: best for performance and research-to-live parity
Best for: experienced developers who want the same engine in backtests and live trading.
NautilusTrader runs Python strategies on a high-performance Rust core, with the same code path for backtests and live trading. A Hyperliquid adapter covers market data and execution.
Pricing: free and open source (LGPL-3.0). A Pro tier and a cloud service are announced but not priced.
Watch-outs: a steep learning curve. Version 2.0 is still a release candidate, and the Hyperliquid adapter is marked as under active development.
6. Jesse: best for clean, readable crypto strategies
Best for: Python developers who want simple crypto strategies and AI help writing them.
Jesse's research side (backtesting and optimisation) is free and open source (MIT), with a local MCP server for Claude Code, Codex, Cursor and others. Live and paper trading, Hyperliquid included, come with a paid plugin.
Pricing: backtesting is free; the live plugin starts at $899 for a one-time Basic licence, and Jesse has announced a move to subscriptions.
Watch-outs: live trading costs extra, and Hyperliquid isn't available as a backtest data source.
7. VectorBT PRO: best for fast backtesting research
Best for: researchers testing thousands of parameter combinations.
VectorBT runs vectorised backtests on NumPy and pandas, which makes huge parameter sweeps fast. PRO adds an MCP server and LLM-ready docs.
Pricing: the open-source vectorbt is free (Apache-2.0 with Commons Clause); PRO is $25/month, $240/year or $500 lifetime.
Watch-outs: research only, with no live execution. Pair it with a broker API or a hosting platform to trade.
8. Alpaca: best broker API for US stocks
Best for: developers building stock, ETF, options or crypto bots in the US.
Alpaca is an API-first broker with Python SDKs, free paper trading with $100k of simulated cash, and an official MCP server for Claude, ChatGPT and other AI tools.
Pricing: commission-free US stocks and ETFs; crypto fees from 0.15% maker and 0.25% taker; market data free, or $99/month for Algo Trader Plus.
Watch-outs: no built-in backtester or hosting, US-focused, and no perpetual futures.
9. Interactive Brokers API: best for global multi-asset reach
Best for: traders who need stocks, options, futures and forex across global markets.
IBKR's TWS API has official Python support, plus a newer Web API, and paper accounts let you test without real money.
Pricing: the API is free; you pay IBKR's trading costs (IBKR Pro Fixed is $0.005 per share with a $1 minimum; IBKR Lite is $0 on US stocks and ETFs).
Watch-outs: TWS or IB Gateway must stay running and logged in, the API is famously clunky, and you host everything yourself.
10. MetaTrader 5 Python package: best for forex and CFD brokers
Best for: forex and CFD traders whose broker offers MT5.
The official MetaTrader5 package connects Python to a running MT5 terminal to read prices and send orders, which reaches hundreds of brokers.
Pricing: free; brokers offer demo accounts for testing.
Watch-outs: Windows only, the terminal must be running, and backtesting stays in MT5's own Strategy Tester (MQL5), not Python.
Why isn't Backtrader on this list?
Backtrader is still widely taught, but its last commit was in April 2023. For new projects, a maintained engine is the safer choice.
Which one should you pick?
- You want your bot running today, on managed infrastructure, with limits outside your code: Antei.
- You want institutional-grade research across many asset classes: QuantConnect.
- You want free, open-source crypto bots and don't mind running a server: Freqtrade, or Hummingbot for market making.
- You only need fast backtests and research: VectorBT PRO.
- You trade through a broker: Alpaca for US stocks, Interactive Brokers for global markets, MetaTrader 5 for forex.
FAQ
Is Python good for trading bots?
Yes. It has the best data and machine-learning libraries, and every major broker and exchange has a Python client. For very high-frequency trading, firms move the hot path to C++ or Rust; most retail and quant strategies never need to.
Can I run a Python trading bot for free?
Yes, with an open-source framework on your own computer or a small server. You then own the uptime, the reconnects and the security. Hosted platforms charge for taking those off your hands.
How do I keep a bot from losing more than I planned?
Put limits outside the strategy code: maximum position size, daily loss and drawdown that the platform enforces even if the code has a bug. Antei enforces them on every deployment.
Can an AI write my Python trading bot?
It can write a good first draft. Backtest it, read what it does, and run it in paper trading before trusting it with money. With Antei, Claude Code or Codex can do the whole loop through MCP.
Try it
Upload a Python strategy, backtest it on Hyperliquid data and run it in paper trading against live prices, with no real money: antei.ai/python.
Antei, ranked #1 here, is our product. Prices were checked in October 2026. Trading involves risk, and automated strategies can lose money. Nothing here is investment advice.
Top comments (0)