Heterogeneous Tail Generalized Common Factor Modeling

A multivariate normal mean-variance heterogeneous tails mixture distribution is proposed for the joint of financial factors and asset returns (referred to as Factor-HGH). The latent variable model incorporates a Cholesky decomposition dispersion matrix ensure rich dependency structure capturing stylized facts data. It generalizes several existing structures, with or without factors. further applicable in large dimensions due fast ECME estimation algorithm all parameters. advantages modelling jointly under non-Gaussian errors are illustrated an empirical comparison study between Factor-HGH classical factor models. While results Fama-French 49 industry portfolios line Gaussian-based models, case highly tail cryptocurrencies, portfolio based on Factor HGH doubles average return while keeping volatility, maximum drawdown, turnover, expected-shortfall at low level.