用人工神经网络方法对股票收益率影响因素的实证分析

An Empirical Analysis on the Influencing Factors of Stock Returns Using Artificial Neural Network

文章用人工神经网络方法对中国股市的股票收益率与有关因素进行了实证分析。用香港理工大学的中国股市数据库(CSMAR)1990-2000数据,采用BP型前馈人工神经网络方法研究了公司规模、交易量、(因子和年收益价格比等四个因素与股票收益率之间的关系。在扩展分析中运用扰动法,并提出影响因子的概念来衡量各参量对股票收益率的影响。计算结果表明,目前对于中国股票市场来说,公司规模对股票收益率具有较强的解释能力,交易量和(因子具有一定解释能力,年收益价格比的解释能力较弱。

This paper uses artificial neural network (ANN) to empirically analyze the influencing factors of stock returns to China's stock markets. Our data for 1990-2000 comes from China Stock Market & Accounting Research (CSMAR) Database of Hong Kong Polytechnic University. The authors use BP Feedforward artificial neural network (ANN) to study the influence of size, trading volume, (3 factor and annual earnings/price (E/P) on stock returns. In expansion analysis, the perturbation method is used and the Impact Factor is proposed to measure the influence of these parameters on stock returns. The calculation results shows that for China stock market at present firm size has relatively strong explanatory power over stock returns, trading volume and β factor have some explanatory power while annual earnings/price (E/P) has relatively weak explanatory power.