摘要
利用数据挖掘手段从网络平台信息提取投资者情绪不仅增加了高频情绪数据的可得性,也有助于深入分析情绪与股票市场运行的互动关系。本文抓取上证指数股吧的实时发帖,通过文本语义分析构建了投资者日内高频情绪指标,并研究了其对股市盘中收益的预测效应。研究发现,中国股票市场的日内投资者情绪能正向预测股票市场运行,这种预测作用下午交易时段表现得更显著;尽管投资者情绪的预测作用独立于收益率自身的盘中动量效应,但在显著性程度上较前期收益率和波动水平要弱;牛市中投资者情绪对日内收益率的预测作用强于滞后收益率等变量,熊市则相反,但在暴涨或暴跌的极端市场环境中,情绪对日内收益率的影响程度相较于滞后收益率等变量更为显著;隔夜投资者情绪的释放会显著影响次日上午的市场收益率,但存在时滞性;午间休市期间的投资者情绪会与上午收益率一起正向影响下午的市场表现;进一步看,噪音交易是投资者情绪影响股票收益率的重要驱动力量。在考虑了月份效应、星期效应以及宏观经济变量的影响后结果仍然稳健。这些结论有助于从更高频率视角深入理解股市中情绪效应的特征及机理。
The method using data mining to extract investor sentiment from network platform information not only increases the availability of high-frequency sentiment index, but also helps to deeply analyze the interaction between investor sentiment and stock market. The paper obtains the real-time posts of Shanghai Securities Composite Index, and constructs the intraday high-frequency sentiment indicators through text analysis, studying its forecasting effect on the stock market intraday returns. The study finds that the intraday investor sentiment can positively predict the stock market return, and the forecasting effect is more pronounced in the afternoon trading session. Although the forecasting effect of investor sentiment is independent from the intraday momentum effect to the stock return itself, it’s less significant than the previous return and volatility. In the bull market, the investor sentiment predicts the intraday return more strongly than the lag return, while the bear market is opposite. In the extreme market environment such as skyrocketing or plunging, the influence of investor sentiment on intraday return is more significant than that of lagging return. The release of overnight investor sentiment will significantly affect the stock return of next morning, but there is a time lag. The investor sentiment during midday break can positively affect market performance in the afternoon along with the stock return in the morning. Further, noise trading is an important driving force for investor sentiment to influence stock return. The results are still robust after considering the effects of the month, the week and macroeconomic variables. These conclusions will help to understand the characteristics and mechanisms of sentiment effects in the stock market from a higher frequency perspective.
作者
尹海员
吴兴颖
YIN Hai-yuan;WU Xing-ying(International Business School of Shaanxi Normal University, Xi’an 710119, China)
出处
《中国工业经济》
CSSCI
北大核心
2019年第8期80-98,共19页
China Industrial Economics
基金
教育部人文社会科学研究一般项目“股票流动性对投资者情绪波动的响应机制研究”（批准号16YJA790061）
中央高校基本科研业务费专项项目“投资者情绪、股票流动性与资产配置效应”（批准号GK201803091）
关键词
高频情绪
股票收益率
日内效应
文本挖掘
high-frequency sentiment
stock return
intraday effect
text mining
