The estimation of multivariate location and scatter is the cornerstone of the classical multivariate statistical methods widely used in portfolio selection problems. However, they are not robust. We propose to use an alternative non-parametric approach based on the weighted L^P depth as robust location and scatter estimator in order to deal with extreme events in asset returns analysis. We first review weighted L^P depth along with its main properties and then discuss its application to portfolio selection through a small simulation study.
Depth-based portfolio selection / Pandolfo, Giuseppe; Iorio, Carmela; D'Ambrosio, Antonio. - (2018), pp. 1-6. (Intervento presentato al convegno SIS 2018 - 49th Scientific Meeting of the Italian Statistical Society tenutosi a Palermo nel 20-22 June 2018).
Depth-based portfolio selection
Giuseppe Pandolfo
;Carmela Iorio;Antonio D'Ambrosio
2018
Abstract
The estimation of multivariate location and scatter is the cornerstone of the classical multivariate statistical methods widely used in portfolio selection problems. However, they are not robust. We propose to use an alternative non-parametric approach based on the weighted L^P depth as robust location and scatter estimator in order to deal with extreme events in asset returns analysis. We first review weighted L^P depth along with its main properties and then discuss its application to portfolio selection through a small simulation study.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
