v2.0.0

Release Notes – v2.0.0

🚀 First Major Release: v2.0.0

We are excited to announce the first major release of VaRpy, a Python library for Value at Risk (VaR) and Conditional Value at Risk (CVaR) modeling using advanced statistical distributions and GARCH models.

---

What's New

• Major Refactoring and Modernization

  • Complete overhaul of the codebase for improved organization, maintainability, and consistency.
  • All VaR models now use a unified, class-based architecture with a shared BaseVar interface.
  • Modular design with private helper methods for key computations and clear separation of concerns.

• Standardized Model Structure

  • Each model (Normal, Student's t, EVT) follows a consistent interface:
    • Constructor only requires theta (confidence level) and horizon (forecast period).
    • .run() method processes return data and computes VaR/CVaR.
    • Consistent naming and comprehensive docstrings across all classes and methods.

• Improved Type Hints and Documentation

  • Extensive use of modern Python type hints (numpy.typing) for clarity and static analysis.
  • All methods and classes are fully documented with clear parameter and return type information.

• Enhanced Code Organization

  • Models are now located in varpy/var/models/ with consistent, lower-case file naming.
  • Project configuration and dependencies are managed via pyproject.toml (PEP 621), with legacy files (setup.py, setup.cfg, requirements.txt) removed for simplicity.

• Flexible and Extensible Design

  • Adding new VaR models is straightforward: inherit from BaseVar, implement run(), and follow the established modular pattern.
  • Comprehensive contributing guidelines provided in the README.

---

Available Models

Model	Description
Normal	Assumes returns are normally distributed; GARCH(1,1) with Gaussian innovations
Student's t	Heavy-tailed returns; GARCH(1,1) with Student's t innovations
EVT	Extreme Value Theory; combines GARCH with Generalized Pareto Distribution for tail modeling

---