Font Size: a A A

A Study On Statistical Evaluation Of Investor Sentiment And Its Application

Posted on:2014-01-14Degree:DoctorType:Dissertation
Country:ChinaCandidate:L CuiFull Text:PDF
GTID:1269330401476719Subject:Statistics
Abstract/Summary:PDF Full Text Request
As of the end of December2012, China has more than2,494listed companies and a large number of these companies have marched towards the path of the modern enterprise by listed restructuring and financing. China’s capital market has not only increased the capacity of the real economy, but also has played an irreplaceable role in shaping the pattern of economic development, speeding up structural adjustment and industrial upgrading, and implementing the innovative national strategy.However, we could not avoid the problem of establishing and updating the assets of oversight mechanisms perfectly. In2007, the subprime crisis has occurred first in the U.S. market, and then evolved into a worldwide recession. Government and international organizations were struggled to cope with it but they also put some actions to summarize and to reflect on it:they made a preliminary agreement, which is the behavior of financial institutions and the pro-cyclicality of the financial system and the lack of systemic risk prevention are the important factors for the global financial crisis. Furthermore, credit rating agencies become the target of public criticism. By analyzing the current rating system, the subjective reason is that rating agencies and rating objects have become a uniformity of different grades from the opposite sites; the object reason is that the rating point of view is mostly based on objective perspective. However, drawback of objective rating is the hysteresis effect of its data sources. Therefore, using both subjective and objective evaluation methods for expanding existing rating points of view, in order to complete the external oversight mechanisms of the relevant stock market is quite necessary and with realistic value.However, the first problem we have to face is from what respect to obtain the subjective data to evaluate the listed company. This paper argues that investor sentiment is the bridge to contact investors, stock market and listed companies. Investor sentiment evaluation method consists of two types:indirect indicators of measurement and a direct indicator of measurement. Indirect measurement must be persuasive, but do not completely reflect investor sentiment, which is to extract the data from financial markets that can indirect reactions proxy variables of investor’s emotion. Direct indicator of measurement can directly reflect the investment by emotions, but vulnerable to the survey cost constraints, because the survey (investors) may also willing to express ideas and ultimately affect the quality investigation. This forces us to turn our attention to the new social media.Blog, micro blogging, social news and networking forum-based social media are filled with the entire Internet space at a rapid pace. In social media, everyone is the publisher of. information, also the disseminator of information, but also the receiver of information. Social media has not only changed the publication and dissemination of information, but also guided the change in the way of people’s social life and investment. Internet evolved from a simple information technology platform to become the main carrier of social media, and develop into a collection of information release, sharing, communication as one of the social network, but also to provide the cheap and abundant data sources about invest sentiment survey. Related studies have pointed out that the network rumors plagued the development of China’s stock market, which has become a chronic problem, not only a serious impact on stock prices, but also greatly influence investment decisions of investors and market confidence. However, because of the internet information is huge, cross discipline theory and the limitation of information technology, makes the ratings of listed companies based on network public opinion repeatedly mention, but there are few empirical studies. Thus, from the investor sentiment index constructed based on the network of public opinion, to explore the relationship between investor sentiment and the stock market, and thus the introduction of public listed companies investor sentiment rating (hereinafter referred to as public rating), become a challenging and full of real value proposition.First of all, with the help of investor sentiment index, we can depth grasp the psychological changes and behavioral characteristics of China’s stock market investors, which can further analysis of the linkage relationship between investor sentiment and stock market. The authority of the investor sentiment index published will help to enhance investor awareness of the market risk, help investors fully grasp the investment value of listed companies, helps to constraints the behavior of the listed companies, help reduce the financial fraud, eventually reduce the stock market information asymmetry. For listed companies, by means of the investor sentiment index can keep abreast of market and investor focus, to make a positive response in a timely manner, to avoid the network rumors adverse impact on the stock market. At present the lack of network public opinion supervision mechanism, leading to excessive spread of false information. Therefore, the significance of this study is to explore ways to enhance the supervision of the stock market network of public opinion, and thus to reduce the systemic risk due to the noise disturbance caused by the stock market.Secondly, the introduction of investor sentiment public rating system is able to provide a reference to improve the existing listed companies rating system. At present, the rating agencies rating of listed companies using objective rating system, and evaluation system focused primarily on financial indicators. Evolution process of company’s financial status and information release delay problem made the evaluation result lags behind. On the other hand, the existing rating systems are powerless facing the listed companies’financial frauds. In addition, the existing rating systems strongly lead into the subjective rating index, but it faces indicators collection and quantifies issues, resulting in a subjective rating index data deviation. To make up for these shortcomings, the emotional tendency of investors based on the network of public opinion analysis, and then the public evaluation of listed companies, will be an effective complement. Rating process text takes feature mining, sentiment analysis and comprehensive evaluation methods to quantify the emotional tendencies of the network text information, to form the expansion of listed companies evaluation perspective.More importantly, the introduction of network public opinion makes it possible to forecast information measure, based on quantifiable investor sentiment index of listed companies to track monitoring investor’s mood changes, and changes in investor sentiment and the linkage relationship between the stock market analysis, to strengthen the management of the stock market in our country and optimize market stability mechanism provides new train of thought and choice, this is under the background of counter-cyclical macro-prudential regulation more practical significance.In this paper, according to "ask questions-analysis problem-problem solving" research train of thought, at the same time, by adopting the combination of statistical evaluation and empirical analysis of economic statistical research paradigm, the empirical research part by combining qualitative analysis with quantitative analysis, as well as the mainly quantitative analysis method on investor sentiment evaluation and application research.First defines the sentiment of this study, combined with core behavioral finance theory to analyze the typical characteristics of the Chinese stock market investor sentiment, focuses on investor sentiment and the network public opinion, the stock market price and rating of the inner link of listed companies, puts forward the hypothesis premise of empirical research and theory in this paper. Secondly, in the existing investor sentiment evaluation method on the basis of comparative analysis, based on network public opinion to build investor sentiment index, use data mining technology to build the technical framework of investor sentiment evaluation model, in Shanghai and Shenzhen300listed companies as the object of measurement, the sample period, the sentiment index of the stock market in our country. Again, the network public opinion to the investor sentiment and the stock market under the linkage relationship between special research, analyses the effect of investor sentiment influence on stock market prices, to verify the theoretical analysis of the three theoretical hypothesis is put forward. Finally, the application of investor sentiment to rating of listed companies, to complement the existing rating system of listed company to provide new ideas.The main results of this article are as follows:(1) Constructions of Investor sentiment index in this study are based on the network public opinion, which can objectively reflect the observation period in our country’s stock market investors’ emotional state, thus, preliminarily verified the text information from the network for investor sentiment index is an effective measurement way.(2) In a variety of methods, on the basis of comparative analysis, this article has verified the emotional dictionary based on the field of financial securities and property thesaurus of fine-grained sentiment analysis method, in the face of massive amounts of text information, emotional tendency quantitative aspects also show certain sensitivity.(3) With the help of investor sentiment index, stock market effect analysis has found that:under the network public opinion, investor sentiment is a systematic factor that affects the stock market, the positive correlation between stock market returns and investor sentiment; there are short-term lag influences on it.(4) Based on text information investors emotions tending to build public rating system of listed companies, listed companies can roughly reflect observation period development status, the analysis has found that macroeconomic policy adjustments and public emergency rating obviously affects the results to the public, to verify the real-time and effectiveness of public rating system.This paper which based on investor sentiment in the network of public opinion in the evaluation of the relevant issues broke through the traditional listed company rating perspective, highlighting the interdisciplinary study of statistical characteristics. Overall, combining of Behavioral Finance, Informational Technology and Statistic Analysis with other theoretical subjects, this article has used network opinions to establish investor emotion data with interpretation of reality, as well as expanding applications of investor emotion data for effects of stock market and the public rating for listed company, in order to investigate new research methodologies for quantifications of complex text data in social science research. According to the result, the most innovative aspects of this article are as follows:Based on behavior finance and statistics index theory, this article has built Investor sentiment index under a network of public opinion. Also with the basic of analysis methods with comparison, it has used combination of characteristics of investor sentiment in public opinions. Seeing the attention indicators as the number of indicators, emotional tendencies as quality indicators, and using quantitative indicators as quality indicators weight coefficient to build investor sentiment evaluation model. This approach has the advantages of simple, intuitive, and with powerful explanation, as well as with comprehensive, correct and timely portrait of changes in stock market investor sentiment. This is not only the expansion of investor sentiment evaluation index source perspective, but also innovative methods of assessment of investor sentiment.According to the methodology of information mining technology and statistical analysis, this article has searched methods to do investor sentiment analysis by using a large number of text information. The use of the network text mining technology including crawl through the web text information, text information pretreatment, feature mining and sentiment analysis to achieve the evaluation index of investor sentiment, extracting investor sentiment from a wide range of unstructured range of information, solving the two key issues of build investor sentiment evaluation:extract attention indicators and quantitative emotional tendencies.First of all, providing a simple and effective treatment ideas for mass text information denoising; second, building the investor sentiment the feature index system and getting properties thesaurus contains632words; third, specifically building an emotional lexicon (containing23,333emotional words) to adapt to the field of securities investment in the process of sentiment analysis. Finally, it has constructed the explanatory power of investor sentiment index which based on the reality of network public opinion and getting the listed company public rating functions. This basic work will help promote a deeper level study on the network of public opinion, the market effect and financial stability mechanism innovation.Using this research to gain investor sentiment data has explored the relationship between investor sentiment and the stock market characteristics index in public opinions. This research has also mainly analyzed the influential effects among the changes of stock market pricing by investor sentiment. Thus, we get three theoretical assumptions:first, the network of public opinion investor sentiment is a systemic factor that affects the stock market rate of return. Second, the network of public opinion under the investor sentiment is a positive impact to the rate of return on the stock market. Third, the impact which investor sentiment in the network of public opinion affects the stock market rate of return is a short-term lag effect, not long-term lag effect.By expanding the application of investor sentiment based on the perspective of the listed companies, extracting rating index system of listed companies from the text information, quantizing investors emotional tendency into the rating index data, thus, we rating the public listed companies, providing a new way of thinking to supplement and improve the existing rating system.In addition, because of the constraints of research level and technical approaches, this article inevitably leaves insufficient issues. Firstly, according to the limit of technical conditions, this study has only selected CSI300constituent stocks of listed companies as research sample, which is not the complete coverage of all listed companies; also the Eastern wealth network shares are selected as the only and typical data source, which could be expanded to micro blogs, blogs, social network sites and other social media platforms to enable a complete analysis. Secondly, regarding the limitation of technical approaches and the scope of knowledge, during the text data processing, while we have established the emotion dictionary and listed property thesaurus in stock investment areas, it could not completely cover the complex emotional vocabulary in network media, which may affect the accuracy of sentiment classification. Therefore actions should be made in future research to build a complete emotion dictionary and listed property thesaurus. Finally, with the limitations of the data processing capability, we have not done further data mining to replying posts, which has lost part of the text data, which can be improved in future research by using in-depth emotion analysis to replying posts in order to gain more reliable emotional information on investors.For further research, this article gives an outlook for in-depth statistics analysis for investor sentiment and applications in order to improve emotional quantitative techniques in network areas with public opinions, solving polarity shift processing problems and considering the influence of the network interference such as "water army" information and so forth.
Keywords/Search Tags:Investor Sentiment, Public opinion of network, StatisticalEvaluation, Text mining, Market effects
PDF Full Text Request
Related items