基于多属性决策和神经网络模型的股价预测研究

Research on Stock Price Prediction Based on Multi-Attribute Decision Making and Neural Network Model

本文选择A股市场中具有代表性的10支股票,根据券商研报所列示的重要特征指标,提取其2011年1月至2021年9月的市盈率、前期涨跌幅、总市值、营业收入增长率等17个特征指标,将其归纳总结,可得到七类特征因子:估值因子、成长因子、盈利能力因子、动量反转因子、交投因子、规模因子和股价波动因子。随后,利用熵权法对特征指标进行赋值,确定各因子权重,依托多属性决策模型中的加权算术平均算子,量化出公司七类特征因子的得分,再探究七类特征因子得分对其对应股票走势的影响,并考虑突发事件的影响,运用自回归模型和神经网络模型建立股价预测模型。

In this paper, we select 10 representative stocks in the A-share market, extract 17 characteristic indicators such as price-to-earnings ratio, previous rise and fall, total market capitalization and operating income growth rate from January 2011 to September 2021 based on the important characteristic indicators listed in brokerage research reports, and summarize them to obtain seven types of characteristic factors: valuation factor, growth factor, profitability factor, momentum reversal factor, trading factor, size factor, and stock price volatility factor. Subsequently, the entropy weighting method is used to assign values to the characteristic indicators, determine the weight of each factor, rely on the weighted arithmetic average operator in the multi-attribute decision model, quantify the scores of the seven types of characteristic factors of the company, and then explore the impact of the scores of the seven types of characteristic factors on their corresponding stock movements, consider the impact of unexpected events, and use the autoregressive model and neural network model to establish a stock price prediction model.