bashtage/arch

What it solves

It provides a comprehensive set of tools for financial econometrics, specifically focusing on modeling volatility and analyzing time-series data to understand risk and price movements in financial markets.

How it works

The library implements various econometric models and tests. It uses Python, with Cython and Numba for performance optimization. Key components include:

• Volatility Modeling: Implements ARCH, GARCH, TARCH, and EGARCH models to estimate how variance changes over time.
• Unit Root and Cointegration Tests: Tools to check if a time series is stationary or if multiple series move together over the long term (e.g., Augmented Dickey-Fuller, Engle-Granger).
• Bootstrapping: Methods to estimate confidence intervals and covariance using IID, Stationary, Circular Block, and Moving Block bootstraps.
• Covariance Estimation: Kernel-based estimators (like Bartlett/Newey-West) for long-run covariance.
• Multiple Comparison Procedures: Tools like the Model Confidence Set (MCS) to evaluate predictive ability across multiple models.

Who it’s for

Financial analysts, quantitative researchers, and econometricians who need to model financial time series and perform rigorous statistical testing on market data.

Highlights

• Support for a wide range of volatility models (GARCH family) and $\text{ARCH}$ models.
• Integrated support for various distributions (Normal, Student's T, Generalized Error Distribution).
• Comprehensive suite of unit root and cointegration tests.
• Performance enhancements via Cython and Numba.
• Built-in bootstrapping tools for robust statistical inference.