NautilusTrader

NautilusTrader can run a strategy against historical quotes, trades, bars or order-book data within one event-driven engine. It is intended for developers who want control over the simulation, including venue balances, fees, data loading and strategy behavior. The project uses the GNU Lesser General Public License version 3.

Build a reproducible test

The low-level BacktestEngine route starts with an engine configuration. Add a venue, instrument, strategy and historical data, then call run() to produce results and performance metrics. The higher-level BacktestNode route uses configuration objects and a data catalog, including chunked loading for larger datasets. Venue fee models must be set explicitly, even for a zero-fee baseline.

Match the venue's order-book setting to the supplied data. Bars cannot recreate market depth or the order of prices within each bar. In a depth-based venue, quotes and bars do not update the matching book, so an unsuitable data choice can leave simulated orders unfilled. Even recorded order-book data cannot show how a hypothetical order would have changed the market.

Match the documentation to the installed version

The current installation guide distinguishes stable 1.x from the 2.x release candidate used by the latest documentation. Stable 1.x cannot run the latest examples unchanged. The project advises against using release candidates for live trading with real capital. Check the selected version's Python and operating-system requirements before installation.

The open-source engine can be installed without a hosted-service account. Live adapters require the relevant venue setup, and external market data or brokerage services may have their own costs. The product site lists a Pro waitlist and a Cloud offering in development, so those should not be treated as generally available hosted access.