基于投资者情绪的股市危机预警系统研究

发布时间：2018-02-02 03:28

本文关键词： 投资者情绪股市危机预警系统 Logit模型 KLR信号分析法　出处：《华南理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：股市危机在当今世界经济一体化和金融自由化的趋势下频繁出现，它所带来的经济成本以及社会成本是异常巨大的，加之我国股票市场发展的特殊性，对股市危机的预警显得更为重要和迫切。本文从行为金融学的视角切入，通过筛选程序选定了上证基金指数、新增基金开户数、上证指数、上证指数成交量和上证A股市盈率为投资者情绪代理变量，使用主成分方法构建投资者情绪。使用单变量Logit模型对沪市在2005年1月至2012年6月之间发生危机的可能性进行了预测。实证结果表明，投资者情绪对股市危机的发生概率有显著的正向影响作用，情绪高涨时触发股市危机的概率大约是情绪低落时发生股市危机概率的7倍，，是预测股市危机的显著预测因子，并且通过了深圳市场的稳定性检验。选出文献中常用的金融危机预警变量，同样使用单变量Logit方法，选出对我国股市危机具有预测作用的宏观变量。实证结果显示，投资者情绪的引入能够提高以常用宏观变量为主的预警模型对股市危机的预测准确度。通过KLR信号分析和阈值遍历方法计算出基于投资者情绪的股市危机预警模型的最佳信号阈值，并统计得本文模型的噪音-信号比率为0.022，信号总体正确率为93.7%，并且样本外测试稳定。说明本文设计的基于投资者情绪的股市危机预警系统对股市危机具有显著的预测作用。最后，本文尝试构建混频投资者情绪和周度投资者情绪并对其预警能力进行了检验，结果显示混频情绪比同频情绪的信息量更大、预警能力更强；周度情绪对股市危机具有显著的中短期预警能力。实证结果为今后构建多时标、多频度的综合性预警系统提供了可行性证明。
[Abstract]:The stock market crisis appears frequently under the trend of economic integration and financial liberalization in the world today. The economic and social costs brought by the crisis are enormous and the particularity of the development of China's stock market. The warning of stock market crisis is more important and urgent. From the perspective of behavioral finance, this paper selects the Shanghai Stock Exchange Fund Index, the number of new fund accounts, the Shanghai Stock Exchange Index, the Shanghai Stock Exchange Index turnover and the Shanghai A-share price / earnings ratio as the proxy variables of investor sentiment through the screening process. The principal component method is used to construct investor sentiment. The univariate Logit model is used to predict the possibility of Shanghai stock market crisis from January 2005 to June 2012. The empirical results show that. Investor sentiment has a significant positive effect on the probability of stock market crisis. The probability of triggering stock market crisis when the mood is high is about 7 times the probability of the stock market crisis when the mood is low. It is a significant predictor of stock market crisis, and has passed the stability test of Shenzhen market. Select the commonly used financial crisis warning variables in the literature, the same use of univariate Logit method, select the macro variables that can predict the stock market crisis in China. The empirical results show that. The introduction of investor sentiment can improve the prediction accuracy of stock market crisis based on common macro variables. The stock market crisis prediction based on investor sentiment can be calculated by KLR signal analysis and threshold traversal method. The optimal signal threshold of the alarm model. The results show that the ratio of noise to signal is 0.022, and the overall correct rate of signal is 93.7%. The results show that the stock market crisis warning system based on investor sentiment has a significant predictive effect on stock market crisis. Finally, this paper tries to construct mixed investor sentiment and Zhou du investor emotion and test its warning ability. The results show that mixed frequency emotion has more information than the same frequency emotion and has stronger warning ability. Zhou du mood has significant ability of short and medium term early warning for stock market crisis. The empirical results provide feasibility proof for the construction of multi time scale and multi frequency comprehensive early warning system in the future.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224

【参考文献】