基于投资者情绪的股市隐马尔科夫模型预测

发布时间：2018-03-29 02:19

本文选题：投资者　切入点：情绪　出处：《华侨大学》2017年硕士论文

【摘要】：行为金融学的研究表明,情绪对投资者的决策有显著的影响,在积极的情绪下,投资者倾向于高估投资机会、低估投资风险,从而交易频繁;在消极的情绪下,则反之。这样的投资行为最终使得股票市场产生了超乎寻常的波动,不利于资本市场的稳健发展。若能从“源头”着手,把投资者的情绪充分运用起来,在研究股票市场的涨跌态势方面将会开创新的思路。本文拟对投资者情绪进行量化,再构建隐马尔科夫模型对股票市场进行预测。本文首先借用情绪-金融决策模型分析了投资者情绪影响股票市场的两条路径:“情绪-认知-行为”路径与“情绪-行为”路径。然后,从理论层面分析了如何以投资者情绪为基础,运用隐马尔科夫模型对股票市场进行预测。在实证分析部分,本文首先对投资者情绪进行测量,运用偏最小二乘法(PLS)构建出了综合指标,该指标表明:消费者信心指数、新增股票投资者开户数、股票交易量和换手率与投资者情绪正相关,封闭式基金折价率与投资者情绪负相关。然后,考虑到我国股票市场“政策市”的实际情况,本文对九次利好政策事件、九次利空政策事件进行了研究,结果表明:政策因素对股票市场的影响已经被包含在了投资者情绪之中,运用投资者情绪对股票市场进行预测无需再单独考虑政策因素的影响。接着,本文根据2007年1月-2015年12月的投资者情绪综合指标数值,计算出了隐马尔科夫模型的关键参数,再根据预测原理运用MATLAB软件实现了对股票市场的预测,结果表明:所预测的时间段越长、期数越多,预测结果出现的概率越小。进一步地,为了突出投资者情绪的作用,本文还运用马尔科夫链进行了预测,并对两种模型的预测结果进行了检验和对比,分析表明:隐马尔科夫模型产生的预测结果更加准确,也就是说,投资者情绪是有助于预测股票市场的。
[Abstract]:Behavioral finance studies have shown that emotions have a significant impact on investors' decisions. In positive emotions, investors tend to overestimate investment opportunities and underestimate investment risks, thus trading frequently; in negative emotions, On the other hand, this kind of investment behavior ultimately makes the stock market have extraordinary fluctuations, which is not conducive to the steady development of the capital market. If we can start from the "source", we can make full use of the investor's sentiment. This paper intends to quantify investor sentiment, which will lead to a new way of thinking in studying the ups and downs of the stock market. Then we construct a hidden Markov model to predict the stock market. Firstly, we use the mood-financial decision model to analyze the two ways that investor emotion affects the stock market: "emotion, cognition and behavior" and "emotion". -behavior "path. Then, This paper analyzes how to predict the stock market with hidden Markov model on the basis of investor sentiment. In the part of empirical analysis, this paper first measures investor sentiment. The synthetic index is constructed by using partial least square method (PLS). The index shows that the index of consumer confidence, the number of new stock investors' account opening, the trading volume of stocks and the turnover rate are positively related to investor sentiment. The discount rate of closed-end funds is negatively correlated with investor sentiment. Then, considering the actual situation of "policy market" in China's stock market, this paper studies nine good policy events and nine bearish policy events. The results show that the influence of the policy factors on the stock market has been included in the investor sentiment, and the influence of the policy factors should not be considered separately when using the investor sentiment to predict the stock market. In this paper, the key parameters of Hidden Markov Model are calculated according to the comprehensive index of investor sentiment from January 2007 to December 2015, and the prediction of stock market is realized by using MATLAB software according to the prediction principle. The results show that the longer the time period, the more the number of periods, the smaller the probability of the predicted results. Further, in order to highlight the role of investor sentiment, this paper also uses Markov chain to predict. The results of the two models are tested and compared. The results show that the hidden Markov model is more accurate, that is, investor sentiment is helpful to predict the stock market.
【学位授予单位】：华侨大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：B842.6;F832.51

【参考文献】