基于个体投资者情感量化的上证指数预测检验

Prediction Test of Shanghai Stock Index Based on Emotional Quantification of Individual Investors

在缺乏验证个体投资者情感量化来预测股指的有效性背景下,对为期三个月的股吧评论数据进行抓取,运用朴素贝叶斯法进行投资者情感的量化分析。在对经过不同期滞后处理的时间序列数据进行相关性分析的基础上,采用VAR模型对上证指数的3分钟均线涨跌幅进行预测,并使用神经网络算法修正预测。结果表明:上证指数涨跌幅与投资者情感之间同时存在线性与非线性关系,且日区间联动幅度较大;构建模型时应综合考虑两者关系和投资者情感的异方差性。

Under the background of the absence of verifying effectiveness on predicting stock index by quantifying individual investor emotion,we grasped comments about the stock performance of the past three months from relevant post bars,utilized Na觙ve Bayes to quantitatively analyze investor emotion,then carried out correlation analysis on the time series data treated in different lag stages; further,used VAR model to predict the index's fluctuation of 3-minute average line,as well as applied neural network algorithm to modify the prediction model. The results show that there are linear and nonlinear relationships between the Shanghai stock index's decline and investor emotion,and the daily linkage is relatively large. Both the relationships and the heteroscedasticity of investor emotion should be considered while constructing model.