Fast LP Algorithms for Portfolio Optimization
22 Pages Posted: 13 Apr 2017
Date Written: April 11, 2017
This paper describes two algorithms for financial portfolio optimization. These algorithms find optimal portfolios for a number of risk measures: CVaR, MAD, LSAD and dispersion CVaR. The algorithms work for discrete distributions of asset returns where optimization problems can be reduced to linear programs. The first algorithm solves the simple recourse problem as described by Klein Haneveld and Van der Vlerk using Benders decomposition method. The second algorithm finds an optimal solution to LP problem with the smallest distance to a given benchmark portfolio and is an adaptation of the least norm solution (called also normal solution) of linear programs due to Zhao and Li. The algorithms are implemented in R in the package PortfolioOptim.
Keywords: Linear Programming, Portfolio Optimization, Asset Allocation, Conditional Value-At-Risk, Mean Absolute Deviation
JEL Classification: C61, C63, G11
Suggested Citation: Suggested Citation