A log-robust optimization approach to portfolio management

摘要: We present a robust optimization approach to portfolio management under uncertainty that builds upon insights gained from the well-known Lognormal model for stock prices, while addressing model's limitations, in particular, issue of fat tails being underestimated Gaussian framework and active debate on correct distribution use. Our approach, which we call Log-robust spirit model, does not require any probabilistic assumption, incorporates randomness continuously compounded rates return by using range forecasts budget uncertainty, thus capturing decision-maker's degree risk aversion through single, intuitive parameter. objective is maximize worst-case value (over set allowable deviations uncertain parameters their nominal values) at end time horizon one-period setting; short sales are allowed. formulate problem as linear programming derive theoretical into optimal allocation. then compare numerical experiments with traditional where applied directly returns. results indicate significantly outperforms benchmark respect 95 or 99% Value-at-Risk. This because leads portfolios far less diversified.