考虑行为特征的多期鲁棒投资组合模型及实证研究

考虑投资者的行为特征以及模型参数的不确定性,构建考虑行为特征的多期鲁棒投资组合模型.在前景理论的基础上,引入动态损失厌恶系数和动态财富参考点,建立动态前景理论价值函数.为了满足投资者的安全性要求,在模型中考虑机会约束,调整模型的保守程度.针对模型多期规划的特点,设计两阶段初始化策略.进一步地,在标准粒子群算法的基础上,根据种群性能的反馈信息,设计多频振动变异操作,提出改进的粒子群算法.实证结果表明:改进的粒子群算法能够有效提高算法的求解精度;考虑行为特征的多期鲁棒投资组合模型能够满足投资者的心理预期,且在实际投资决策中具有可行性.

Empirical study of multi-period robust portfolio model with behavioral character

Considering investors' behavioral factor and the uncertainty of parameters, a multi-period robust portfolio model with behavioral character is developed. Based on the prospect theory, we propose a dynamic prospect theory value function, where both of the loss aversion parameter and reference wealth are updated dynamically. In order to satisfy investors' safety requirement, chance constraint is introduced into the portfolio model, which enables flexibly adjusting the degree of conservatism of the solution. Based on the characteristic of multi-period planning in the portfolio model, a two-stage initialization strategy is designed. We also present an improved particle swarm optimization with multi-frequency vibrational mutation operator, which considers the feedback information during the searching process. The empirical results show that the algorithm improves the precision of the solution, as well as the proposed model meets the investors' psychological expectation and is viable in practice.