基于新闻文本情绪的区间值股票回报预测研究

Forecasting Interval Valued Stock Returns Based on News Media Sentiments

投资者情绪与股票市场的价格变动息息相关,所以正确理解投资者情绪对金融投资者的投资策略选择与监管部门的风险管控具有重要意义.本文选取国务院新闻文本与金融情感词典,首先构建一个基于粉丝加权的新闻媒体情绪区间指数,然后建立自回归条件区间模型,对我国股市主板市场与不同板块子市场进行预测研究.实证研究结果表明:1)基于粉丝加权的新闻媒体情绪指数能够有效地反映新闻情绪,且对我国股票市场回报产生显著的负向影响;2)不论在主板市场还是不同板块子市场,粉丝加权的新闻媒体情绪指数对我国股市回报的预测效果显著最优,为文本情感指标的构建提供了新的思路与实践;3)新闻媒体情绪对主板市场股票回报的预测效果随着预测步长的增加而减弱,但对不同板块子市场股票回报的预测效果具有异质性.研究表明基于文本挖掘的新闻媒体情绪区间指数对股市区间值回报预测具有重要作用.

Since the investor sentiment is closely related to the price movement of Chinese stock market,it is crucial to correctly understand the investor sentiment for both financial investors and regulators for risk management.This paper aims to construct a fan-weighted news sentiment index based on the State Council news texts and financial sentiments dictionary,and then to investigate the predictability of the fan-weighted news sentiment indicator on the main board market and submarkets of Chinese stock market by an interval-valued autoregressive model.The empirical results show that:1)Our proposed fan-weighted news sentiment index can not only reflect news sentiment effectively,but also has a significant negative impact on Chinese stock market;2)Compared with the simple average news sentiment index,our proposed fan-weighted news sentiment index can significantly improve the predictive ability on Chinese stock market under both the main board market and sub-markets,which provides a new idea for constructing sentiment indicators;3)The predictive ability of news sentiment index gradually decreases over time for the main board market,but it is heterogeneous for the sub-markets.These empirical findings show that the investor sentiment mined from news text is important to predict the stock market performance.