Predicting equity returns using Twitter sentiment
Abstract
Predicting events and responding to unforeseen events quickly has been a significant struggle. Previous publications have shown the ability to monitor and predict events using semantic analysis. These publications have been able to use semantic analysis to predict movements in the stock market and respond to events faster than ever before. This thesis aims to fuse these ideas to develop and test a new approach to semantic scoring and test its effectiveness. The semantic scoring focuses on mood related word referencing a single company, a sector, and the market for predicting stock returns. In addition to returns, focus was placed on predicting direction of the market as well. This work found an inconclusive relationship between sentiment and the returns over the following days. With the framework from this thesis in place, a re fined wordlist and modeling could improve predictive accuracy.
URI
http://purl.galileo.usg.edu/uga_etd/muhlheim_mitchell_d_201305_mshttp://hdl.handle.net/10724/28837