E1 - News Sensitive Stock Trend Prediction

From the stock market to the commodities market, Hong Kong's financial markets provide the opportunity for everyone to be involved in the local and international economy. This project focuses on predicting the impacts of the mass media on the movement of stock prices. In contrast to the traditional time series analysis, where predictions are made based solely on the historical performance of the time series, here, predictions are made according to the non-quantifiable information – news stories. Our novel time series and text mining techniques allow us to figure out the inherent dynamics that drive movements in the prices of financial assets by news stories.

Applications

• Decision support for on-line trading
• Acts as an advisor for both fund managers and individual investors
• Real-time solvability
• Suitable for many applications related to concurrent time series and text mining, such as sales vs advertisement analysis

Features

• A new research area that combines two interrelated but different subjects
• Predict stock trends based on non-quantifiable information – news stories
• Uncover the hidden yet valuable information for bank decisions and marketing investors
• Continuous-time framework, reflecting the dynamic nature of the asset market

Website
http://www.se.cuhk.edu.hk/~dbgroup

Principal Investigators
Prof. Jeffrey Yu / Prof. Wai Lam
Department of Systems Engineering and Engineering Management