EECS 6893 Columbia F19 Project Our primary goal is to accurately predict the stock price of individual companies based on historical price and Twitter sentiment. Our approach is different than previous results we are aware of because of the scope and scale of our analysis. We gathered an entire year of tweets and independently tested four sentiment analysis tools and tested a wide range of models. This resulted in a prediction model that is more accurate than the results we are aware of and has a higher potential to generalize to real-time and long-distance price forecasting. This example uses Nike data, but the approach can be applied to any publicly traded company with a large social following.