NautilusTrader Review in 2026
9 min read

NautilusTrader is a free, LGPL v3 licensed trading engine with a Rust core and Python API that runs backtests and live trading on the same strategy code. It suits small systematic teams with engineers who want to own their stack. It is single node, has no open source UI, and its v2 API is still a release candidate.
What this covers
- NautilusTrader had 31,337 GitHub stars on November 7, 2026 and is maintained by Nautech Systems.
- It is licensed under LGPL v3, which permits commercial use and lets proprietary strategies built on an unmodified engine stay private.
- Backtest, sandbox and live modes share strategy code, but the docs list six ways live behavior can differ.
- The open source scope is single node; dashboards, distributed backtesting and built in ML tooling are out of scope.
- The engine is free; data, infrastructure and engineering time are the real costs, and Pro pricing is not published.
- Latest stable is v1.231.0; the v2 line is a release candidate and v1 gets only critical security fixes.
NautilusTrader is a free, open source trading engine with a Rust core and a Python strategy API. It lets one team backtest and run live trading strategies on the same code. It is a strong fit for a small quant team that has engineers and trades liquid markets. It is not a finished product for a fund or family office that wants dashboards, managed data and support without building anything.
This review covers what the project is, what it actually does, what the LGPL license allows, what it costs to run, and where it fits for a fund, family office or deal team. Every product fact below comes from the project's GitHub repository, its documentation and website, PyPI, or independent listings, all fetched on November 7, 2026. The repository is at github.com/nautechsystems/nautilus_trader.
NautilusTrader at a glance
| Item | What the sources say |
|---|---|
| What it is | An open source, Rust native engine for multi asset, multi venue trading systems, with Python as the control plane |
| Maintainer | Nautech Systems (listed as the PyPI author) |
| GitHub stars and forks | 31,337 stars and 4,230 forks on November 7, 2026 |
| Project age | Repository created in June 2018 |
| License | GNU LGPL v3 (GitHub lists LGPL-3.0; PyPI lists LGPL-3.0-or-later) |
| Latest stable release | v1.231.0, published August 2, 2026 |
| Next major line | v2 release candidate (v2.0.0rc8 on the website install snippet), Python 3.13 or 3.14 |
| Modes | Backtest, sandbox (live data, simulated fills) and live |
| Stated scope | Single node backtesting and live trading for individual and small team quant traders |
| Paid layer | NautilusTrader Pro: server, dashboard and a hosted MCP service. No prices on the Pro page |

AI for finance teams
Screen more deals without adding analysts
First system live in about six weeks. You own the code.
Book a discovery callWhat NautilusTrader is
The project's own description is a "production-grade, Rust-native engine for multi-asset, multi-venue trading systems." The design goal is to remove a common split in quant work. Research usually happens in vectorized Python, and the production system is then rewritten in a compiled language. That rewrite is where backtest results and live results drift apart.
NautilusTrader runs research, simulation and live execution inside one event driven architecture. The core is written in Rust. Python bindings are provided through PyO3, so strategy logic, configuration and orchestration can stay in Python. Teams that want everything compiled can write the whole system in Rust.
The website reports more than 3.3 million downloads and more than 6,000 Discord members. GitHub shows one dominant contributor, cjdsellers, with roughly 19,780 contributions, followed by a long tail of contributors in the low hundreds. That pattern is common in serious open source infrastructure, and it is worth knowing if you are judging key person risk.
What it actually does
Backtesting
The backtester runs multiple venues, instruments and strategies at the same time. It accepts historical quotes, trades, bars, order books and custom data at nanosecond resolution. The documentation includes separate sections on fill models, bar execution, fill prices and matching, and accounts and margin, which is where most backtest realism problems live.
Live and sandbox trading
The documentation lists three use cases: backtest on historical data, sandbox with real time data and virtual execution, and live trading on real or paper accounts. The same strategy and execution algorithm code runs in all three. Optional state persistence is backed by Redis or PostgreSQL.
Orders and execution
The domain model supports time in force options including IOC, FOK, GTC, GTD, DAY, AT_THE_OPEN and AT_THE_CLOSE. It also supports post only, reduce only, iceberg orders, conditional triggers and OCO, OUO and OTO contingency orders. The docs note that venue support varies by adapter, so a given order type is only as available as the venue allows.
Integrations
Venues and data providers connect through adapters. The README lists 21 integrations, including Interactive Brokers, Binance, Bybit, Coinbase, Kraken, OKX, Deribit, dYdX, Hyperliquid, Betfair, Polymarket, Kalshi, and data providers Databento, Tardis, CoinAPI and FinFeed. Each adapter carries a status of planned, building, beta or stable. An independent directory, Open Source Alternatives, counts 17 stable integrations. Check the status of the specific adapter you need before you plan around it.
The coverage leans toward crypto venues, with Interactive Brokers as the main path into traditional equities, futures, FX and options. Any venue with a REST API or WebSocket feed can in principle get an adapter, but the roadmap says the project has limited bandwidth to accept new official integrations.
Security and release hygiene
The README lists an unusually detailed supply chain posture. It includes pinned and checksummed Rust dependencies, cargo-vet and cargo-deny audits, SLSA build provenance on Python artifacts, Sigstore signed container images with SBOMs, and signed commits on protected branches. For a fund whose security team reviews open source before approval, this list shortens that conversation.
The license and what it allows commercially
NautilusTrader is licensed under the GNU Lesser General Public License, version 3. According to the Choose a License summary, the LGPLv3 permits commercial use, modification, distribution, patent use and private use. Its conditions are that you make available the source of the licensed library and any modifications to it, under the same license, and keep the copyright and license notices.
The important part for a firm is the library exception. A larger work that uses the library through its interfaces may be distributed under different terms, without releasing the source of that larger work. In practice, your proprietary strategies sitting on top of an unmodified engine do not become open source. If you modify the engine itself and distribute it, those engine changes carry the LGPL obligations.
One detail to raise with counsel: GitHub labels the license LGPL-3.0, the README's dependency policy refers to LGPL-3.0-only, and PyPI labels the package LGPL-3.0-or-later. These labels differ on whether later versions of the license apply. It is a small point, but a legal review should note which text governs.

Build library
See what we have already shipped
Working systems, the stack behind them, and what they cost to run.
Open the libraryWhat it costs to run
The engine is free. The real costs are infrastructure, market data, engineering time and, if you want it, the commercial Pro layer.
Infrastructure
The open source scope is single node. One process runs a node, with optional Redis or PostgreSQL for state. The project ships Docker images on GitHub's container registry, including a JupyterLab image for research. Supported platforms are Linux on x86_64 and ARM64, macOS on ARM64 and Windows on x86_64. Distributed backtesting orchestration is explicitly out of scope, so large parameter sweeps need your own scheduler.
Market data
NautilusTrader does not bundle data. It provides adapters to data providers such as Databento, Tardis, CoinAPI and FinFeed, and to venue feeds. Those providers set their own terms, and their pricing was not verified for this post. For many teams, historical tick and order book data is the largest recurring cost line, so budget it before you commit.
Model API costs
There are none in the engine. The roadmap lists integrated hyperparameter optimization and built in AI or ML tooling as out of scope. The website and docs do say the engine is fast enough to train AI trading agents with reinforcement learning or evolutionary strategies, but you bring your own frameworks and compute. If you add a large language model to your workflow, that cost sits outside NautilusTrader.
NautilusTrader Pro
Nautech Systems sells a commercial layer called NautilusTrader Pro. According to the Pro page, it includes a server that runs one live node and exposes a control plane API, a dashboard to monitor and control nodes, and a hosted MCP service that gives development tools semantic search over NautilusTrader code and documentation. The server and dashboard are delivered as Docker containers on your infrastructure. The page states that these components send no telemetry or trading data back to Nautech.
The Pro page marks the dashboard image as "under development" and says the number of node servers you can connect depends on your plan. It does not publish prices. The website also links a cloud platform page, which did not load during research for this post, so this review makes no claims about it.
Limits to be honest about
The API is still moving. The README warns that the project is under active development, that some features may be incomplete, and that breaking changes can occur between releases. The project aims for a release every two weeks. The v2 line is still a release candidate, and v1 now receives only critical security backports. A team starting today should start on v2 and expect migration work.
Backtest and live are not identical, and the docs say so. Shared strategy code is not the same as identical behavior. The live trading documentation lists six sources of difference: venue rules, transport failures that leave an order outcome unknown, timing and message interleaving, persistence, orders created outside the node, and reconciliation with venue reports. That candor is a good sign. It also means live deployment is real engineering, not a configuration switch.
No user interface in the open source project. UI dashboards and frontends are out of scope by policy. If a portfolio manager or investment committee needs to see positions and risk on a screen, you either build that, buy Pro once its dashboard ships, or wire the engine into your existing tools.
Single node by design. The scope is built for individuals and small teams. A multi desk fund with central risk and many strategies will need orchestration, monitoring and data infrastructure that the project leaves to the user.
Who should use it and who should not
Good fit: a small systematic team with Python and ideally some Rust skill, trading crypto, futures, FX or equities through supported venues, that wants backtest and live on one codebase and wants to own the stack. The Open Source Alternatives listing frames it as a self hosted alternative to QuantConnect for teams that want control over data, engine and strategy IP.
Poor fit: a team with no engineers, a team that wants a managed environment with built in data, or a team whose need is portfolio reporting rather than trade execution. The same independent listing notes that a managed platform is faster to start with when running infrastructure is not acceptable.

Implementation services
Put AI into the workflow you actually run
Diligence, close, reconciliation, reporting — in your environment.
See how it worksWhere it fits for a fund, family office or deal team
Hedge funds and prop desks with a systematic book. This is the natural home. The engine covers research, simulation and execution, and the LGPL lets proprietary strategies stay private. Plan for data costs, an engineer who owns the stack, and a deliberate move to v2.
Family offices. Most single and multi family offices do not run high frequency or market making strategies. If yours runs a systematic sleeve with an in house quant, NautilusTrader is a credible engine for it. If the real problem is consolidated reporting across managers, custodians and private positions, a trading engine is the wrong tool. Our notes on AI for family offices cover the workflows that usually matter more.
Private equity and deal teams. NautilusTrader does not touch sourcing, diligence, CIMs or portfolio monitoring. A PE deal team gets little from it directly. The place it can show up is a fund's liquid markets or treasury function. For deal work, see AI for private equity and our AI deal desk systems.
When an in environment build is the better answer
NautilusTrader is a good engine. It is not a system. The pieces it leaves out on purpose, such as dashboards, distributed runs, data pipelines, monitoring and the reports an investment committee reads, are often where a fund spends most of its effort. That gap is where many firms decide whether to assemble open source themselves or have it built.
That is the work consultance.ai does. We build custom finance AI inside the client's own environment, with the first system live in about six weeks, and the client owns the code. Sometimes that means wrapping an engine like NautilusTrader with the data, controls and reporting a fund needs. Sometimes it means a different tool entirely. If you are weighing that choice, you can book a call or browse the free workflows in our library.
Sources
- NautilusTrader GitHub repository, README and API metadata, fetched November 7, 2026
- NautilusTrader roadmap and open source scope, fetched November 7, 2026
- NautilusTrader live trading docs, backtest and live differences, fetched November 7, 2026
- NautilusTrader installation guide, fetched November 7, 2026
- NautilusTrader website, fetched November 7, 2026
- NautilusTrader documentation overview, fetched November 7, 2026
- NautilusTrader Pro page, fetched November 7, 2026
- nautilus_trader on PyPI, fetched November 7, 2026
- Open Source Alternatives listing for nautilus_trader, fetched November 7, 2026
- Henry Lin, Introduction to NautilusTrader, DEV Community, fetched November 7, 2026
- Choose a License, GNU LGPLv3 summary, fetched November 7, 2026
