The increasing globalization and interdependence of financial markets have motivated the development of multivariate models for the volatility and covariance structures of financial assets. Accurate forecasting of these structures is crucial for financial risk management and informed decision-making. A central challenge is to develop parameterizations that ensure positive definiteness while remaining sufficiently parsimonious to avoid the curse of dimensionality.
The first part of the talk will review the main approaches proposed in the literature, starting with classical multivariate extensions of GARCH-type models. Particular attention will be devoted to alternative parameterizations and statistical methods designed to reduce the number of parameters while retaining flexibility.
The widespread availability of high-frequency financial data has subsequently led to the development of realized measures of volatility. In particular, realized covariance matrices provide ex-post estimates of the daily covariance structure of asset returns and offer a natural basis for dynamic covariance modelling.
The second part of the talk will focus on recent models for realized covariance matrices, discussing their advantages and limitations. We will then present some new parameterizations designed to achieve a better balance between parsimony and flexibility, while preserving the positive definiteness of the covariance matrices.
2 Ottobre 2026, ore 12
Edoardo Otranto
Dipartimento di Scienze Sociali ed Economiche - Sapienza Università di Roma
In person: Room 34 (4th floor) building CU002 Scienze Statistiche
Webinar: https://uniroma1.zoom.us/j/83625004899?pwd=bXCtz0mp759PUh2lkqT0BUoVa0Uegg.1
ID riunione: 836 2500 4899
Passcode: 123456