基于Bi-LSTM模型挖掘的股吧投资者情绪对股价泡沫的影响  

Investor Sentiment Mining Based on Bi-LSTM Modeland Its Impact on Stock Price Bubbles

作　　者：尹海员[1] 杨庆松 YIN Haiyuan;YANG Qingsong(Shaanxi Normal University,Xi'an,China)

机构地区：[1]陕西师范大学国际商学院

出　　处：《管理学报》2022年第12期1874-1885,共12页Chinese Journal of Management

基　　金：教育部人文社会科学研究规划基金资助项目(22XJA790008)。

摘　　要：为探究投资者情绪对股价泡沫的影响与预测能力,搭建双向长短时记忆神经网络模型,对东方财富股吧中样本票的实时发帖文本进行情感识别与分类,构建投资者情绪日度指标;进一步地,运用GSADF检验法检验股价泡沫,并实证研究投资者情绪与股价泡沫之间的关系。研究表明:投资者情绪显著正向影响个股泡沫的存在概率及泡沫强度;这种影响在小规模、低股权集中度和非国有企业中更加显著;中介效应检验发现,投资者情绪会通过影响股票交易金额对股价泡沫产生影响。In order to explore the influence and prediction ability of investor sentiment on stock price bubble, this study constructs a two-way long-short-term memory model to identify the sentiment polarity of real-time posting texts of sample stocks, and constructs the daily investor sentiment index. Further, based on the General Sup ADF method, the stock price bubble is tested and the relationship between investor sentiment and stock price bubble is empirically studied. The empirical results show that investor sentiment has a positive effect on the probability and strength of stock bubble;This effect is more significant in small-scale, low equity concentration and non-state-owned enterprises;The mediation test shows that investor sentiment has an impact on stock price bubble through current transaction amount of the stock.

关 键 词：投资者情绪 股价泡沫 Bi-LSTM GSADF检验法 

分 类 号：C93[经济管理—管理学]