ON PORTFOLIO SELECTION UNDER EXTREME RISK MEASURE: THE HEAVY-TAILED ICA MODEL
Abstract
This paper is devoted to the application of the Independent Component Analysis (ICA) methodology to the problem of selecting portfolio strategies, so as to provide against extremal movements in financial markets. A specific ICA model for describing the extreme fluctuations of asset prices is introduced, stipulating that the distributions of the ICs are heavy tailed (i.e., with power law behavior at infinity). An inference method based on conditional maximum likelihood estimation is proposed for our model, which permits to determine practically optimal investment strategies with respect to extreme risk. Empirical studies based on this modeling are carried out to illustrate our approach.