Reversion Strategy for Online Portfolio Selection with Transaction Costs

Abstract

This paper is about the online portfolio selection problem with transaction costs, which is an unavoidable factor in real financial trading. By exploiting the mean reversion property of stock prices, we propose a portfolio selection strategy named 'mean reversion strategy with transaction costs (MRTC)'. To avoid overmuch transaction costs, the strategy adaptively transfers a proper amount of capital between stocks to adjust the turnover. Furthermore, we conduct numerical experiments on several real market datasets, and show that our proposed algorithm outperforms the existing state-of-the-art ones when taking transaction costs into account.