【作者】 冯丽娜；

【导师】 张永杰；

【作者基本信息】 天津大学 ， 管理科学与工程， 2014， 硕士

【摘要】 随着互联网的快速发展，股票市场是对信息技术和信息传播应用最为普遍的领域之一。结合当前中国互联网发展现状，人们主要关注了国家宏观政策导向、股票本身信息量和股票论坛、博客等各种社交网络媒体传递信息的内容，对于集合了各种信息渠道的搜索引擎，是人们最为常用的信息获得和传播的工具。本文主要采用了宏观经济数据（SHIBOR利率）、百度新闻指数，利用数据挖掘和计量经济学等方法，就开源信息对股票市场行为的影响进行了分析。通过一系列的研究结果表明：（1）不论是在SHIBOR利率上升时期还是在波动时期，当模型中加入成交量后，个股的波动性明显减小，说明成交量可以有效解释股票市场的价格波动。并且利率波动时期，信息传递能更好的解释股票市场的价格波动。（2）信息传递对条件波动性具有积极的影响。当加入互联网开源信息时，波动持续性下降的幅度最为明显（从0.9850至0.3564），而加入原始成交量和成交量的调整值时，波动持续性的下降程度并没有加入互联网信息的明显。这些结果意味着，互联网信息是信息传递更好的代理。（3）在分时间段的具体研究中，通过非参数相关检验证明，四个时间段的信息到达的代理变量与该段时间的abs（异常收益率）具有显著的非参数相关关系。第一时间段，也就是上午开市前的异常收益率变化最为剧烈，市场反应程度较大。上下午交易时间的异常收益率较为一致，而下午开市前的反应程度较低。在另一方面，本文还发现除第一时间段外，其他三个时间段的异常收益率都是之后时间段内信息传递的Granger结果，因此说明了存在发布信息之前的信息泄漏问题。本文结合了两种信息传播渠道，进行高频数据挖掘，既有利于发现日内交易等行为结构，也可以为深入挖掘信息传播和股价波动两者之间关系提供新的分析思路，具有良好的理论价值和应用价值。更多还原

【Abstract】 With the rapid development of the Internet, the stock market has widely appliedinformation technology and information dissemination. Combined with the currentsituation of China’s Internet development, the main concern is the nationalmacro-policy orientation, the stock information and the message delivered by stockforum, blog and other social media. A search engine has collected various ofinformation channels and become the most used tools to obtain information andcommunicate.This paper has used the macroeconomic data (SHIBOR rates), Baidu news index.Through data mining and econometric methods, it can analyze the impact of opensource information on the stock market behavior. Through a series of research, resultsshow that:(1) Whether in the period of fluctuations or rising of SHIBOR, when add thetrading volume into model, the volatility has significantly reduced, indicating thattrading volume can effectively explain the price fluctuations of the stock market.Besides, during the fluctuations of interest rate，information transmission can betterexplain the price fluctuations of the stock market.(2) Information transmission has a positive impact on conditional volatility.When joining the Internet open source information, the volatility persistence is mostdecreased (from0.9850to0.3564), while joining the original trading volume andadjustment values of trading volume, the reduction of the volatility persistence wasnot as obvious as the Internet information. These results mean that, the Internetinformation is the better agent of information transmission.(3) In the concrete research, by nonparametric relevant certificate, it can provethat four period Baidu news and the corresponding abs (abnormal yield) hassignificant nonparametric correlation. For the first period, that is, the abnormal returnsbefore the start of stock market own the most dramatic change, market reaction degreeis bigger. An abnormal return in morning trading is more uniform in afternoon trading,while the reaction in period3is low. On the other hand, this paper also found thatexcept for the first period, the other three periods after period of abnormal returns are Granger result of information transmission, thus illustrates the problem of informationleakage before releasing information.This paper has combined the two channels of information dissemination. Afterhigh frequency data mining, it not only can find intra-day trading behavior, but alsoprovide new train of thought to dig deeper into relationship between informationdissemination and stock price fluctuations. Therefore, it has the good theory value andapplication value.更多还原