Title: Penalized enhanced portfolio replication with asymmetric deviation measures
Authors: Gabriele Torri - University of Bergamo (Italy) [presenting]
Rosella Giacometti - Università di Bergamo (Italy)
Sandra Paterlini - University of Trento (Italy)
Abstract: The problem of enhanced portfolio replication is addressed. A strategy is proposed based on the minimization of novel risk deviation measures based on expectiles. Such risk measures allow accounting asymmetrically for the differences between the portfolio and the benchmark, favouring positive deviations compared to negative ones. The model nests the minimum TEV replication approach as one special case. Ridge and elastic-net regularization penalties are added to the model to control estimation error better and improve the out-of-sample performances. Simulations and real-world analyses on multiple datasets allow us to discuss the pros and cons of the different methods.