基于神经网络集成理论,建立股市预测模型.其中分别建立"基本数据模型"、"技术指标模型"和"宏观分析模型",最后以简单平均生成集成系统.实证分析表明,股市预测神经网络集成系统的泛化能力高于各个独立的模型,从而使模型具有更好的稳健性和更好的应用价值.

The technique of artificial neural networks provides a novel and effective method for stock market forecast. The neural network ensemble can heighten the generalization. In this paper, we proved that the generalization of stock market forecast system based on neural network ensemble is superior to the single models and the system is more effective and applicable.