基于投资者行为参数的股票指数广义回归神经网络预测模型

General Regression Neural Network Forecasting Model of Stock Index Based on Behavioral Parameters of Investors

在运用神经网络模型对股票价格进行短期预测时,一般的神经网络预测模型都是以价格的时间序列滞后作为输入变量,但是由于影响价格的因素错综复杂,很多因素无法准确测量,而且市场信息的噪音太大,因此预测效果往往不太理想,于是如何选择有效的输入变量就成为一个困扰这项研究的难题。

When making short - term forecasting for stock price, general neural network models take lags of price time series as input variables. But the factors affecting price are complex, and many are impossible to be accurately measured and the noise of market information is too heavy, so forecasting effect is not satisfactory. Then how to select effective input variables becomes a disturbing problem in research.