Online Portfolio Selection Models versus Mean Variance Optimal Choices

In this paper we discuss two different online portfolio selection models and we compare their performance with some classic take and hold mean variance portfolio selection strategies. In particular, we examine two online portfolio selection models (the Passive Aggressive Mean Reversion model (PAMR) proposed by Li, Zhao, Hoi and Gopalkrishnan (2012) and the Online Moving Average Reversion model (OLMAR) proposed by Li and Hoi (2012)) both based on the mean reversion characteristics of the returns. Therefore, we propose an empirical analysis that compares their performance on the US stock market with the classic mean variance approach. Finally, we discuss the advantages and limits of the proposed portfolio selection strategies.