Previsione del Value at Risk e dell'Expected Shortfall: Analisi Comparativa di Modelli Predittivi Classici, Econometrici e di Reti Neurali

This master’s thesis investigates the fundamental role of risk management in financial decision-making, focusing on the computation of Value at Risk (VaR) and Expected Shortfall (ES) as essential tools for quantifying and mitigating financial risk. The study encompasses a comprehensive exploration and comparison of classical methodologies, including historical simulations and Monte Carlo simulations, with cutting-edge econometric models such as the GARCH model and emerging neural network-based approaches. The introductory section of the thesis lays the foundation by providing a thorough understanding of financial risk concepts, along with clear definitions of VaR and ES. Classical methods for VaR and ES calculation are presented, emphasizing historical simulations and Monte Carlo simulations. Subsequently, the thesis introduces econometric models and delves into the architecture and training process of neural networks for VaR and ES computation. To evaluate the performance of these diverse methods, extensive experiments are conducted using real financial data spanning various market conditions. Rigorous evaluation metrics are employed to assess the accuracy and reliability of each approach in capturing extreme market events. The findings highlight the promising capabilities of econometric and neural network models in capturing complex risk patterns, particularly during periods of heightened market volatility and non-linear dependencies, surpassing traditional methods. In-depth discussions interpret the results, providing insights into the strengths and limitations of each approach. This research adds valuable contributions to the ongoing discourse on risk management techniques by emphasizing the advantages of employing neural networks for VaR and ES calculations. It underscores the significance of incorporating innovative machine learning approaches in financial risk management practices, offering guidance for financial professionals and policymakers in navigating the ever-evolving and uncertain financial landscape.