🏛️ BenQ-Core

Ben Quant Core is an end-to-end quantitative research framework designed for high-performance stock analysis. It combines robust data engineering with unsupervised deep learning to identify market regimes, model asset behavior, and develop data-driven strategies aimed at achieving consistent alpha.

📂 System Architecture

The codebase follows a modular architecture designed for scalability, separating concerns between data ingestion, alpha generation (modeling), and strategy execution.

.
├── documentation/       # Technical manuals and module instructions
└── src/
    ├── src_DD/          # Data Ingestion & Safety Layer (ETL)
    ├── engine/          # Core Analytics & Prediction Engine
    │   ├── src_features/    # Factor Engineering (Raw/Robust/Neutral)
    │   └── context/         # Market Regime Detection (Unsupervised Learning)
    ├── strategy/        # Strategy Development & Backtesting Framework
    └── dashboard/       # Front-end & Visualization (In Development)

🏗️ Steps of Benq-V1

At the moment: Phase 2

• Phase 1: Data Infrastructure

  • Fault-tolerant ETL pipeline using yfinance with exponential backoff.
  • Data quality and integrity control (split detection, gap filling, etc.) via safety.py.
  • OLAP storage layer based on DuckDB and Parquet.
  • Culminates in the function: MarketLoader().

• Phase 2: Modeling System Development

  • Features: Module dedicated to feature engineering, culminating in the function: get_feature_matrix().
  • Context: Advanced module for Market Regime Detection.
  • Mini-models: Development of specialized models following a Mixture of Experts (MoE) architecture.
  • Meta-Model: Integration layer that processes inputs from this phase to generate a global prediction.
  • Safety Engine: Security module to verify data integrity and calculation accuracy, including automated error logging.

• Phase 3: Strategy & Backtesting - Development of optimized trading strategies based on the Meta-Model's output, supported by a rigorous and high-performance backtesting system.

• Phase 4: Dashboard & Production - Development of a production-ready front-end with a professional visual interface.

  • Integration of a dedicated module to connect the strategies defined in Phase 3 with major Broker APIs for live execution.

---

Detailed documentation can be found in BENQ-CORE/documentation

🚀 Fast start

Requirements

pip install -r requirements.txt