RegimeFlow

RegimeFlow is a regime-aware quantitative trading stack for backtesting, live execution, and research automation. It combines a C++20 engine, Python bindings, regime detection workflows, live broker adapters, risk controls, and a dynamic plugin system across Linux, Windows, and macOS.

Documentation: https://gregorian-09.github.io/regime-flow/

Choose Your Path

If you want to...	Start here
Install the Python package and run a backtest quickly	docs/getting-started/quick-install.md
Build RegimeFlow from source	docs/getting-started/installation.md
Run the first source-based backtest	docs/getting-started/quickstart.md
Use the Python API and CLI	python/README.md and docs/python/overview.md
Inspect live-trading boundaries and broker support	docs/live/overview.md and docs/live/brokers.md
Consume RegimeFlow from CMake or vcpkg	docs/getting-started/quick-install.md
Contribute code or plugins	CONTRIBUTING.md

Features

• HMM-based and plugin-driven regime detection, including transformer-oriented plugin examples.
• Event-driven backtesting with walk-forward optimization, slippage, transaction-cost modeling, and risk overlays.
• Live adapters for Alpaca, Binance, and Interactive Brokers when built with ENABLE_IBAPI=ON.
• Dynamic plugins for strategies, regime detectors, risk managers, and execution models across .so, .dylib, and .dll targets.
• Python bindings and a Python CLI for research workflows, reporting, and dashboard generation.
• Packaging for wheels, Linux .deb/.rpm artifacts, and vcpkg overlay/custom-registry consumption.

Platform Support

Platform	Compiler	Status
Linux x86_64	GCC 10+, Clang	CI
Windows x64	MSVC 2022	CI
macOS arm64	Apple Clang	CI
macOS x86_64	Apple Clang	CI

The canonical install and support matrix lives in docs/getting-started/installation.md.

Installation

Python wheels

pip install regimeflow
regimeflow-backtest --help

Source build

cmake -S . -B build
cmake --build build --target all
ctest --test-dir build --output-on-failure

CMake or vcpkg consumer

find_package(RegimeFlow CONFIG REQUIRED)
target_link_libraries(your_target PRIVATE RegimeFlow::regimeflow_engine)

Detailed install paths, support tiers, and feature flags are in docs/getting-started/installation.md and docs/getting-started/quick-install.md.

Quick Example

import regimeflow as rf

cfg = rf.BacktestConfig.from_yaml("examples/backtest_basic/config.yaml")
engine = rf.BacktestEngine(cfg)
results = engine.run("moving_average_cross")

print(rf.analysis.performance_summary(results))
rf.analysis.write_report_html(results, "report.html")

Running Tests

ctest --test-dir build --output-on-failure
pytest python/tests

Repository Layout

• docs/ user guides, tutorials, architecture notes, API reference, and reports
• examples/ end-to-end configs, Python workflows, and plugin examples
• include/ public C++ headers
• src/ C++ libraries and CLI tools
• python/ Python package, bindings, and Python-specific docs
• tests/ C++ and Python tests
• packaging/ Debian, RPM, and vcpkg consumer assets
• ports/ vcpkg overlay/custom registry port definition

Contributing

See CONTRIBUTING.md.

License

Released under the MIT License. See LICENSE.