A Parallel Integer Relative Robust Mean-RCVaR Model for Portfolio optimization

摘要

Portfolio optimization models are widely adopted in asset management, quantitative trading, and other applications. Relative robust portfolio optimization further considers the situation that the optimization result of the absolute robust optimization model only depends on the worst case. To apply the relative robust portfolio model to inseparable assets, this paper proposes an integer relative robust optimization model based on mixed-integer programming. The experimental results show that the integer relative robust portfolio model can achieve a higher rate of return, lower relative risk, and superior balance between robustness and profitability. Furthermore, to deal with massive computing loads of the model when applied to large-scale assets and largescale historical data, a parallel version of the integer relative robust optimization model is implemented with MPI, that can achieve excellent speedup ratio and scalability.