The financial mood and confidence indexes should be more rapid, comprehensive, and cost-effective than existing surveybased indexes. Based on social media feeds, web search terms and query volumes, and online financial news/blogs, this RP will (i) develop big text and data mining algorithms to identify real-time financial market mood and confidence indexes; (ii) develop visualisation technique to allow identified mood and confidence indexes to be openly accessible to both financial institutes and public. The innovation is the first real-time text mining technique to model dynamic behaviours of financial mood and confidence indexes. This research project is partly based on our recent paper (Bollen, J., H. Mao, X. Zeng, 2011), which used twitter mood to predict stock prices.
We expect to show that online identified indexes are more reliable, accurate and informative by comparing correlations and prediction accuracies to financial market movement. The correlations and prediction analysis should provide earlier insights/ warnings of market movements, risks, and crises for financial markets. Visualisation and corresponding index data can assist financial organisations and policy makers in market analysis and risk assessment. For each index, it will be identified and retrieved in real-time the corresponding web data resources (e.g. market and business news) and blogs and social media data (e.g. twitter).
Moreover, real-time big data/text mining methods will be designed and implemented to identify financial market mood and confidence indexes. Visualisation to display real-time series of mood and confidence indexes and make accessible to financial institutes and the public will be developed.
Early Stage Resercher working on the project: Rytis Simanaitis
