Overview

RegimeFlow is a C++ core with Python bindings for regime-aware backtesting and live execution. The system is built around a single event pipeline that produces consistent behavior across backtest and live modes.

What It Provides

• Data ingestion, validation, and canonicalization.
• Regime detection and regime-aware analytics.
• Strategy execution with configurable costs and risk limits.
• Live trading via broker adapters.
• Plugin architecture for detectors, strategies, data sources, and execution models.

System Map

flowchart LR
  subgraph Data
    A[CSV/Tick/OrderBook] --> B[Data Sources]
    B --> C[Validation + Normalization]
    C --> D[Bar Builder]
  end

  subgraph Engine
    D --> E[Event Generator]
    E --> F[Event Loop]
    F --> G[Strategy Context]
    G --> H[Strategy]
    H --> I[Execution Pipeline]
    I --> J[Portfolio + Metrics]
  end

  subgraph Regime
    D --> R1[Feature Extractor]
    R1 --> R2[Regime Detector]
    R2 --> H
    R2 --> J
  end

  subgraph Live
    L1[Broker Adapter] --> F
    F --> L1
    F --> L2[Message Bus]
  end

Core Guarantees

• One strategy contract for both backtests and live trading.
• One event pipeline for bars, ticks, and order books.
• One configuration surface across C++ and Python formats.

Where To Go Next

• explanation/architecture.md
• explanation/data-flow.md
• explanation/event-model.md