This repository is dedicated to a comprehensive project conducted by Group O, focusing on Value at Risk (VaR) prediction for multi-asset investment portfolios. The study delves into various models including Copula, GARCH, and traditional methodologies, contextualized against the backdrop of major global financial incidents like the European debt crisis and the COVID-19 outbreak.

The project aims to construct and analyze an equally weighted portfolio comprising diverse asset classes. The focus is on the SPDR S&P 500 ETF Trust (SPY), iShares 20+ Year Treasury Bond ETF (TLT), and SPDR Gold ETF, each chosen for their relevance to significant financial events and inherent characteristics.

• GARCH Model: Introduced by Engle and Bollerslev, the GARCH model is utilized to understand the lagged variances in financial data.
• Gaussian/student-t Copula: Employing copula models to understand dependencies between different asset classes.
• Monte Carlo Simulation: Used for forecasting portfolio VaR and back-testing the model's performance.
• Traditional Methods: Including historical simulation and the variance-covariance method.

The data, sourced from Bloomberg Terminal, includes assets like TLT, gold, and SPY. The study involves processing this data, calculating log-returns, and understanding their statistical properties such as skewness and kurtosis.

The project presents detailed results from the GARCH specification, copula analysis, Monte Carlo simulation, and traditional methods. It concludes with insights into the best-fitting models for each asset and the overall portfolio.

The study provides a nuanced understanding of VaR in multi-asset investment portfolios, highlighting the efficacy of combining GARCH, Copula, and traditional methods for risk prediction and management.

A comprehensive list of references is included, offering a depth of context and background to the methodologies and theories applied in this project.

Additional resources and supplementary materials related to the project are provided in this section.

This repository serves as a detailed documentation and source of code for the project "VaR Prediction for Multi-Asset Investment Portfolios Based on Copula, GARCH, and Traditional Models" by Group O.