A0286
Title: Sparse graphical modelling for minimum variance portfolios
Authors: Riccardo Riccobello - University of Trento (Italy)
Giovanni Bonaccolto - University of Enna Kore (Italy) [presenting]
Philipp Johannes Kremer - EBS Universitaet fuer Wirtschaft und Recht (Germany)
Sandra Paterlini - University of Trento (Italy)
Malgorzata Bogdan - Lund University (Sweden)
Abstract: Graphical models have demonstrated exceptional performance in uncovering the conditional dependence structure among a given set of variables. Two novel graphical modeling techniques are introduced: Gslope and Tslope, which use the Sorted L1-Penalized Estimator (Slope) to directly estimate the inverse of the covariance matrix. We develop ad hoc algorithms to efficiently solve the underlying optimization problems: the Alternating Direction Method of Multipliers for Gslope and the Expectation-Maximization (EM) algorithm for Tslope. The methods are suitable for both Gaussian and non-Gaussian distributed data and take into account the empirically observed distributional characteristics of asset returns. Through extensive simulation analysis and real-world applications, we demonstrate the superiority of our new methods over state-of-the-art covariance matrix estimation techniques, particularly regarding clustering and stability characteristics.
