中国股票市场的趋势预测——基于多维交互验证法

Stock Market Trend Prediction:Based on Multi-Dimensional Cross-Validation Method

股票市场是现代金融的核心组成部分,准确有效地预测其运行规律,对于投资者和管理者来说都极为重要。尽管大量研究表明,股票市场是一个高度复杂的非线性动态系统,其变化规律具有一定的趋势性,但如何预测与确认其趋势一直都是困扰投资者的难题。本文提出了多维交互验证法,尝试从驱动面、心理面、行为面等多个维度出发,来预测股票市场的趋势。本文还针对上证综合指数建立多维决策树模型,对该方法进行检验。实证分析的结果表明:多维交互验证法在预测股票市场的月度趋势方面,具有较好的有效性和可行性。

The stock market is the key component to the modem finance. Forecasting its trend accurately and effectively, is of primary significance. Many researchers have already found that the stock market is a highly complicated nonlinear dynamic system, of which certain trend does exist. Nonetheless, how to predict the trend and make use of it remains an unsolved problem. The authors come up with a new method, multi-dimension cross-validation approach, to predict the stock market＇s trend, from the aspects of driving factors, psychological factors and behavioral factors. In the end, we build a statistical decision tree model to predict the Shanghai Composite Index＇s trend to testify our new method. The result shows that, our model performs very well with historical data, and it is feasible and reliable to employ our approach to forecast the market＇s trend.