Reamer Research from Reamer Labs is a Python backtesting library for results that will decide where real money goes. It is a Python and C++ backtesting and live trading engine for strategies on bars of any type: time, tick, volume, dollar, imbalance or your own rule. Costs are modelled per instrument, a written specification states how every order fills, the same inputs give byte-identical results, and its relay sends the same orders live. When a run ends, its visual trade diagnostics tool charts each closed trade on its own bars, beside every value your strategy printed, so you can see why each trade happened. It is a free download, and $20 a month removes the Evaluation Mode limits. If you only need a quick first look at an idea, a free library may be all you need, and the table below shows which.
Python backtesting libraries compared
Releases, licences and prices were checked against each project's own pages and PyPI on 16 November 2026. Reamer Labs publishes this page and sells Reamer Research.
| Library | Licence | Best for | Several instruments | Latest release | Live trading | Price |
|---|---|---|---|---|---|---|
| Reamer Research | Commercial | Strategies on bars with modelled costs, written fill rules and every trade charted | Yes | 4.11.3, November 2026 | Its relay, to your broker through the open-source receiver | Free download in Evaluation Mode; $20 a month |
| backtesting.py | AGPL-3.0 | One strategy on one instrument, with charts | No | 0.6.6, July 2026 | Not covered in its documentation | Free |
| Backtrader | GPL-3.0 | Learning, from years of tutorials | Yes | 1.9.78.123, April 2023 | Older connectors | Free |
| bt | MIT | Portfolio weights and rebalancing | Yes | 1.4.1, October 2026 | None advertised | Free |
| vectorbt | Apache 2.0 with Commons Clause | Large parameter sweeps on arrays | Yes | 1.2.1, November 2026 | Signals only | Free |
| vectorbt PRO | Commercial | vectorbt with more tools | Yes | Private repository | Signals only | $25 a month, $225 a year or $500 lifetime |
| NautilusTrader | LGPL-3.0 | Ticks, order books, many venues | Yes | 1.231.0, August 2026; 2.0 in release candidates | Venue adapters | Free |
| LEAN | Apache 2.0 | Many asset classes, QuantConnect's engine | Yes | LEAN CLI 1.0.233, October 2026 | Self-run, with supported brokers | Free |
| QuantRocket | Commercial | Data, Moonshot and Zipline backtesters and brokers in one platform in Docker | Yes | Client 2.12.0.4, October 2026 | Interactive Brokers, Alpaca and others | Paid; prices shown after login |
| Zipline Reloaded | Apache 2.0 | Existing Zipline code | Yes | 3.2.1, October 2026 | None advertised | Free |
| Freqtrade | GPL-3.0 | Crypto trading bots | Yes | 2026.10, October 2026 | Crypto exchanges | Free |
Reamer Research
- Bars of any type. Time, tick, volume, dollar, imbalance or your own rule, from memory or from memory-mapped files larger than RAM. Each instrument can have one bar per second.
- Costs and fills you can read. Spread, slippage, commission, overnight swap and margin are set per instrument and are part of every run. The execution specification states how orders fill, including a stop and a target inside the same bar.
- Repeatable. With the same data, settings and seed, output is byte-identical, including randomised spread and slippage.
- Live with the same orders.
reamer_relay_*sends the orders your strategy returns to your broker through the open-source receiver over the published protocol and reads the fills back as position state. See How do I take a strategy from backtest to live trading without a rewrite? - Every trade, charted. The visual trade diagnostics tool shows each closed trade on its own bars, with entry, exit, stop and target marked, every indicator or signal your strategy printed drawn as a series beside it, the console lines for those bars, the equity curve and a Monte Carlo simulation of the run.
open_viewer=Trueopens it when a backtest ends. It is one HTML file with no install and no network. - Fast from Python. In the benchmark below, it ran 500,000 bars in 3.38 s from Python and 0.36 s from C++.
- Free to download, with public examples. The kit downloads free with no key or sign-up and runs in Evaluation Mode: one ticker per backtest, up to 30 days or 50,000 bars, no live trading. Example strategies, templates and data-loading scripts are public under the MIT License at github.com/reamerlabs/reamer-research-examples.
- Your data, your indicators. Bars come from any source, or from your Alpaca or Databento account through the bundled fetchers, and indicators are your own code or any Python library. It runs on Linux x86-64, macOS on Apple Silicon and Windows through WSL2.
The other libraries
- backtesting.py is the simplest start: a small API, interactive charts and a built-in optimizer, one instrument per strategy. See What are the best backtesting.py alternatives in 2026?
- Backtrader has the most tutorials and a simple
next()method per bar, but no release since April 2023. See What should I use instead of Backtrader now that it is no longer updated? - bt builds strategies from allocation blocks, which suits rebalancing more than order-level rules.
- vectorbt tests large grids of parameters on whole arrays. Rules that depend on the current position or on bracket orders are harder to express. See What is a vectorbt or vectorbt PRO alternative for event-driven backtesting?
- NautilusTrader has a Rust core and Python strategies, and the same code runs in backtest and live, down to order books. It takes more setup, and its README warns that breaking changes can occur between releases.
- LEAN is the engine behind QuantConnect, run on your own machine with your own data, with strategies in Python or C#.
- QuantRocket is a paid Python platform in Docker with the Moonshot and Zipline backtesters, data services and connections to Interactive Brokers and Alpaca. See What are the best QuantRocket alternatives in 2026?
- Zipline Reloaded keeps Zipline code running. Its maintainers point new projects to ML4T Backtest.
- Freqtrade is a crypto trading bot with backtesting built in.
What a published benchmark measured
What a Backtest Costs ran four engines on the same 20-bar breakout over 500,000 bars, on one AMD EPYC 9575F on 2026-10-25, with no commission, spread or slippage.
| Engine | Time |
|---|---|
| Reamer Research from C++ | 0.36 s |
| Reamer Research from Python | 3.38 s |
| NautilusTrader 1.231.0, one function patched | 13.61 s |
| Backtrader | 15.04 s |
| LEAN, built from source | 22.10 s |
| NautilusTrader 1.231.0, stock | 76.85 s |
The study's author builds and sells Reamer Research, so the runners and raw results are public. It is one machine, one date, one instrument and two simple strategies, and it did not test QuantRocket, backtesting.py, bt, vectorbt, Zipline or Freqtrade. Your own indicators and any model inside the loop count in every engine.
How to choose
| If you want | Look at |
|---|---|
| Results you can trust and repeat, every trade charted with what your strategy saw, and the same orders live | Reamer Research |
| A free first look at one instrument | backtesting.py |
| Portfolio weights and rebalancing | bt |
| Large parameter sweeps in Python | vectorbt or vectorbt PRO |
| Ticks, order books or ready venue adapters | NautilusTrader |
| Many asset classes | LEAN |
| Data, backtesters and brokers in one self-hosted platform | QuantRocket |
| A crypto bot | Freqtrade |
Reamer Research costs $20 per seat per month, cancel any time. See pricing.
Reamer Research is also a free download, with no key or sign-up. Without a licence it runs in Evaluation Mode: one ticker per backtest, up to 30 days or 50,000 bars, and no live trading. A licence removes those limits, and the same code runs unchanged. Example strategies, strategy templates and data-loading scripts for CSV, .bin, Parquet, pandas, Alpaca and Databento are public under the MIT License at github.com/reamerlabs/reamer-research-examples.