Regime switching and outliers in range-based volatility models

• Grant number: 25/20266-2
• Support Opportunities: Scholarships in Brazil - Master
• Start date: October 01, 2025
• End date: February 28, 2027
• Field of knowledge: Applied Social Sciences - Economics - Quantitative Methods Applied to Economics
• Principal Investigator: Carlos Cesar Trucios Maza
• Grantee: Cristiane Yi Jung Li
• Host Institution: Instituto de Matemática, Estatística e Computação Científica (IMECC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 25/06131-7 - Modelling and Forecasting risk measures with intraday data: daily and high-frequency approaches, AP.R

Abstract

Modeling and forecasting volatility, as well as other risk measures, play a central role in numerous economic and financial applications, including portfolio allocation, risk management, option pricing, and trading strategies. Among the most widely used approaches are the GARCH family of models, which assume that volatility at time $t$ is fully determined by its own past values and past squared returns. In recent years, several studies have shown that incorporating intraday data into such models can substantially enhance volatility forecasts and, consequently, the forecasting of other risk measures such as Value-at-Risk (VaR) and Expected Shortfall (ES). In particular, the use of accessible intraday data (open, close, high, and low prices) has emerged as a promising alternative, outperforming traditional GARCH models and delivering performance comparable to variants based on high-frequency data, but at considerably lower computational and financial costs. Despite these advantages, GARCH models relying on accessible intraday data generally fail to account for two stylized facts frequently observed in financial time series: the presence of additive outliers and regime shifts. Addressing these aspects constitutes the central focus of this project. The expected outcomes include the publication of a research article and, ideally, recognition in the form of Best Student Paper Awards at international scientific conferences. Furthermore, the methodologies developed will be implemented and released in public GitHub repositories and packaged in an R library to be submitted to CRAN, ensuring that end users can directly benefit from the advances generated by this research.