Long and Short Term Risk Control for Online Portfolio Selection

Online portfolio selection is to allocate the capital among a set of assets to maximize cumulative returns. Most of online portfolio selection algorithms focus on maximizing returns without effectively controlling risk of loss in return. Further, many risk control algorithms use the maximum drawdown, the Sharpe ratio, and others as risk indicators. However, these risk indicators are not sensitive to the short-term of loss in return. This paper proposes the Long and Short Term Risk (LSTR) control algorithm for online portfolio selection. LSTR achieves high return and low risk by combining the effects of two parameters. The first parameter learns the long-term risk of the market, and its posterior probability changes slowly according to the mean reversion theory. The second parameter senses the short-term risk of the market and makes a quick response to changes in short-term returns. Through the multiplication of the two parameters, the risk control ability of online portfolio selection is effectively improved. The experimental results of the six datasets demonstrate that the performance of LSTR is better than the online portfolio selection algorithms with risk control and those without risk control.