A Parallel Hybrid Intelligent Algorithm for Fuzzy Mean-CVaR Portfolio Model

Chen Li, Zhonghua Lu, Yonghong Hu, Fang Liu, Jue wang

Abstract:

In recent years, fuzzy portfolio selection theory has been well developed and widely applied. Based on the credibility theory, several fuzzy portfolio selection models have been proposed. The fuzzy Mean-CVaR portfolio model is one of the state-of-the-art. However, its’ fuzzy nature which increases the computational complexity makes it take a long time to solve. In order to solve the fuzzy Mean-CVaR portfolio model efficiently, a hybrid intelligent algorithm is designed by integrating Genetic Algorithm (GA) with adaptive penalty function, Simulated
Annealing Resilient Back Propagation (SARPROP) neural network and fuzzy simulation techniques, and to accelerate the computation speed further, we parallelize the hybrid intelligent algorithm with MPI technology. In order to demonstrate its validity and efficiency, we achieve numerical experiments on the Era supercomputer, and the results are compared with the method which is obtained by integrating traditional GA and fuzzy simulation directly. The results show that hybrid intelligent algorithm can get better performance. Experiments under
different processor cores also achieved on the Era supercomputer demonstrate the scalability of the parallel hybrid intelligent algorithm.