THESIS
2014
xi, 58, 3 pages : illustrations ; 30 cm
Abstract
Reviewing the last few financial crises in human history, we found that the assets' price, such
as stock price or housing price, were much higher than their intrinsic value before the crises,
and much lower than intrinsic value during the crises. We know from these experiences and
past studies that public investment sentiment, in addition to intrinsic value, plays a significant
role in asset pricing. However, few researches have been done in studying the influence public
sentiment has on asset pricing because it is hard to measure quantitatively, and we can't study
what we can't measure. Along with the rapid development of Internet media and Web 2.0,
people begin to discuss stock markets on the Internet, especially the social network such as
blogs, Facebook and twitter, which ge...[
Read more ]
Reviewing the last few financial crises in human history, we found that the assets' price, such
as stock price or housing price, were much higher than their intrinsic value before the crises,
and much lower than intrinsic value during the crises. We know from these experiences and
past studies that public investment sentiment, in addition to intrinsic value, plays a significant
role in asset pricing. However, few researches have been done in studying the influence public
sentiment has on asset pricing because it is hard to measure quantitatively, and we can't study
what we can't measure. Along with the rapid development of Internet media and Web 2.0,
people begin to discuss stock markets on the Internet, especially the social network such as
blogs, Facebook and twitter, which generates lots of information containing public opinions
towards the market shown in text form. What I aim to do in these thesis is to derive the public
sentiment toward future stock market quantitatively based on these large-scale text data.
Because of the huge amount of these data and the fact that all these data are in text form, I
developed an automatic financial text classifier by the text classification techniques to extract
quantified public sentiment signals. After study the relationship between the sentiment data
and broad market index, I found that the sentiment signals and market returns interact with
each other closely. Sentiment signals do help predict the market returns, and market returns
also have a big influence on future sentiments.
Post a Comment