This is an Information Retrieval project carried out in collaboration with David Sierra Gonzalez and Xuyang Wu. The project attempts to predict the US stock market based on analysing Twitter status updates. We make use of deep learning sentiment analysis models, ridge regression and (sparse) lasso regression, and find that, yes, popular stocks can be predicted to a certain extend by looking at the volume of tweets concerning them.

Prediction of changes in the stock market using twitter and sentiment analysis, Research Paper (PDF)

The code is also available on GitHub.