NautilusTrader

Production-grade trading engine. Rust-native. Open source. Deterministic event-driven core. Research-to-live parity.

The same backtest config runs live with no rewrite: backtest the GridMarketMaker on Tardis data, then point the same parameters at a Rust LiveNode with the deadman's switch armed. The adapter also fans order submits across parallel connections for redundancy.

When we add a venue adapter, we look for primitives that make live trading safer to run. BitMEX gives the engine two it uses directly: a server-side cancel-all timer that clears your orders if the client goes dark, and redundant order submission across parallel connections.

New tutorial: grid market making on @BitMEX XBTUSD. Backtest the shipped strategy on free Tardis quote data, then run the same config live in Rust. Built around the deadman's switch: a server-side cancel-all timer the engine refreshes on a schedule, so a dropped connection clears your stranded quotes. nautilustrader.io/docs/lates…

NautilusTrader just crossed 1.5M downloads. 269k in the last 30 days alone. Thank you to everyone building with us. The best is ahead. pepy.tech/projects/nautilus-…

On-chain volume attributed to NautilusTrader across @HyperliquidX and @Polymarket now over $100 million. $56 million in the past two weeks. These are 2 of the 15 venues we support. Two more DEX venues land soon. Live stats: nautilustrader.io/resources/…

The engine is open source by design. You and your auditors can read its logic and trace what it does, rather than trust a black box. The higher the stakes, the more transparency matters. github.com/nautechsystems/na…

When an engine touches live capital, you should be able to prove what you're running, not take it on trust. NautilusTrader ships reproducible, signed builds with SLSA Build Level 3 provenance. Verify any release yourself. nautilustrader.io/security/s…

Information-driven bars sample by content rather than clock time. A quiet hour might generate no new bars; a burst at the London open can generate many. @lopezdeprado spent chapters on why it gives better-behaved inputs for systematic work.

The @krakenpro adapter used in the tutorial is first-class in the engine and implemented in @rustlang, same interface as every other venue adapter: nautilustrader.io/docs/night…

Information-driven De Prado bars, VPIN, and Hurst on Kraken. End-to-end out of the box available now, and the same code runs in backtest and live: nautilustrader.io/docs/night…

The clock in NautilusTrader does more than return timestamps. It defines how time enters the engine: as monotonic nanosecond timestamps, as scheduled events, and as a dependency shared across backtesting and live trading. This allows timer-driven logic to run against historical data and real markets through the same interface and execution model. nautilustrader.io/blog/clock…

NautilusTrader is a deterministic, event-driven trading engine that runs one execution model across research and production. Why NautilusTrader exists, the architectural decisions behind it, and why that matters in practice: nautilustrader.io/blog/why-n…

Further architectural and deployment notes will follow.

The system is designed for multi-venue deployment. Venue adapters operate at the edge of the core runtime. The execution model remains consistent across venues.

The architecture is event-driven. Communication flows through a central message bus. Data, execution, risk, and portfolio operate within a unified runtime.

The same core system operates in both research and live environments. Execution behavior is consistent. Time is modeled deterministically. State transitions are explicit.