CMStatistics 2022: Start Registration
View Submission - CFE
Title: Conditional autoregressive G model for common factor detection in the stock market Authors:  Marco Girardi - University of Padova (Italy) [presenting]
Massimiliano Caporin - University of Padova (Italy)
Abstract: A new distribution is presented to model financial assets' realized covariances, obtained as the product of a scalar component distributed as a unit-mean inverse gamma and a matrix component following a Wishart distribution. The mean of the resulting distribution is endowed with an autoregressive moving average structure. The model captures a common factor in the assets' behaviour, which constitutes the inherent risk in the market, as well as the idiosyncratic risk component. The one-step-ahead forecasts of the covariance matrix are employed in a portfolio allocation framework with the aim of tracking the reference index performance by limiting the impact of specific risks. A possible empirical application of the model in a hybrid portfolio management strategy is also discussed.