摘要
Dynamic risk management requires the risk measures to adapt to information at different times, such that this dynamic framework takes into account the time consistency of risk measures interrelated at different times. Therefore, dynamic risk measures for processes can be identified as risk measures for random variables on an appropriate product space. This paper proposes a wavelet feature decomposing algorithm based on the discrete wavelet transform that optimally decomposes the time-consistent features from the product space. This approach allows us to generalize the multiple-stage risk measures of value at risk and conditional value at risk for the feature-decomposed processes, and implement them into portfolio selection using high-frequency data of U.S. DJIA stocks. The overall empirical results confirm that our proposed method significantly improves the performance of dynamic risk assessment and portfolio selection.