The Effect Of Media On Stock Markets

Asset price fluctuation and media information publishing, dissemination and absorption are closely related. Since most investors are unable to directly observe the operation of Listed Corporations, the media has become an important channel for their information access. However, the media is not a cold tool; journalists, editors and audience are all human beings. Under the assumption of bounded rational investors, the mechanism of media influence on asset price can be simply described as "media-investors-asset price". Despite the news content is rich, existing research methods usually condense "all information into a point", also known as "One Point Method". For example, OLS linear regression transforms the media coverage to a factor, as one of the many explanatory variables. However, in the era of big data, by "One Point Method", it is difficult to deal with effects of mass media information on asset price, the key lies in two points:First, rich conent in media suffer a serious loss when they are transformed into One Point; Second, linear regression model cannot capture complex correlation between massive information and asset price. Hence, this study attempts to solve two problems in the view of methodology:First, this study quantilize median content from "One Point Method" to N-dimensional vector by extracting noun, sentiment words in news media. This will reduce the information loss caused by "One Point Method"; Second, this study utilizes support vector regression ("SVR" for short) to capture the complex relation between media content and asset price. Furthermore, this work also discusses the application of media quantization in securities investment and market regulation. Specifically, this study mainly includes the following four parts:First, quantization of media information. As media theorist Mcluhan Marshall said, the media is the expansion of human feelings, it is part of the message itself. There the media is not only the carrier of information, but also the carrier of hu(?)motion. The traditional "One Point Method" condenses rich media information into "one point", which suffers a serious loss.In this study, noun words are extracted as the representative of the "fundamental information"; the psychology words are extracted as the representative of the sentiment information and then media content is successfully transformed into an N-dimensional vector. In order to accurately extract sentiment words in the media content, this study constructs the first Chinese Financial Sentiment Dictionary (CFSD, which includes 1390 positive words and 1890 negative words.Second, analysis of complex relationship between the media and asset price. At present, there are three different models to simulate the relationship between media content and asset price:statistical model, regression model of econometrics, data mining model in the field of artificial intelligence. The variable model focuses on the relationship between media content and asset price, and the econometric model pays great attention on the causal relationship between them, and the data mining model focuses on the complex relationship between them. With the development of the internet society, the complexity of the media information increases, and the data mining model that fits for complex nonlinear relationship has obvious advantages. This study puts forward the "Media-Aware Quantitative Technology", namely "MAQT". Based on SVR, MAQT capture both N-dimensional media content and stock information, tries to establish a complex nonlinear relationship between the media content and future stock price. MAQT studies the relationship between media content and asset price from the perspective of complex relationship, which is a useful supplement to the statistical model and econometric regression model. The results from MAQT show that the media content can best match the price 26 minutes after the news released, therefore,26 minutes can be treated as the best time window to disclose the media effect on asset price.Third, the application of MAQT in securities investment, In this part, the sensitivity of MAQT is deeply analyzed, such as the analysis of MAQT on different media content, different industries and different characteristics of the company. Based on MAQT, this part constructs a conditional MAQT strategy, which outperforms other traditional investment strategies.Fifth, market regulation on media abuse in stock market. With the development of the internet, the media (news, analysts report, internet forum) have the unprecedented influence on stock market. When stock market in its downturn, some investors use malicious media spread bad news and complete the short transaction, known as "malicious short sell", In order to detect illegal clues in malicious short sell, this work proposes the media aware monitoring method, In addition, this study also gives some management suggestions from legal aspects.