Meng, XiaochunAlmowelhi, Almothana2023-11-132023-11-132023-09-12https://hdl.handle.net/20.500.14154/69671In the intricate domain of financial risk management, accurately forecasting Value at Risk (VaR) and Expected Shortfall (ES) using high-frequency data emerges as a present-day challenge. This research aims to explore the accuracy of VaR and ES forecasts across varying high-frequency data intervals. To do so, it leans on the Functional Autoregressive Value-at-Risk (FARVaR) model, a semi-parametric approach introduced by Cai et al. (2019) that leverages the richness of intraday returns. The study outlines the potential of applying this model to high-frequency trading data from leading stock indices such as FTSE_100, DAX_30, and S&P_500. Furthermore, it proposes rigorous backtesting methods, including ECP, BPZ, and the CC and DQ tests, against well-known VaR models like HS, GARCH, and FHS. While this research presents a comprehensive plan for empirical analysis, it does not offer executed empirical results or findings. Thus, it serves as a groundwork for future studies in this realm and underscores the significance of VaR and ES in the contemporary financial landscape.18enValue-at-RiskExpected Shortfallhigh-frequency dataThesis