基于集成预测的均值-方差-熵的模糊投资组合选择

通过基于集成预测的方法构建模糊投资组合, 代表性地选择了遗传神经网络模型、 多因素SVM回归模型和ARIMA时间序列模型作为组合预测中的单一模型, 并将单一模型预测结果作为模糊变量进行投资组合优化, 实证结果表明基于集成预测的均值 -方差-熵的投资组合相比其他组合收益率更高, 相对风险更低. 该方法可以用于投资基金管理、 金融风险管理等实际工作中,以便提高决策的科学性.

Mean-variance-entropy fuzzy portfolio selection based on integrated forecast

In this paper a fuzzy portfolio selection optimization decision based on a combination of forecasting model is constructed, representative selection of the genetic neural network model, multi-factor SVM regression model and ARIMA time series model as a combination of a single model for forecasting. This paper establishes possibilistic mean-variance-entropy portfolio selection model in which a single model to predict the result is treated as fuzzy variable, the empirical results show that the possibilistic mean-variance-entropy portfolio selection model can obtain higher yield and lower relative risk. The methodology of this paper is useful in such practice as fund management, financial risk management. It is hoped that the methodology can improve the scientific level of decision making.