Understanding the Effect of Investors’ Sentiments on the S&P 500 Price Levels : A Deep Learning Model Approach

This paper aims to examine to what extent can investors’ sentiments extracted from social media content, specifically Twitter, improve the predictability of the S&P 500 price levels. Two Recurrent Neural Network models were built; the first one is solely based on historical records and technical indicators. The second one includes the same variables as the first model, along with the outputs of the sentiment analysis, performed using TextBlob library. While assessing the performance of both models created, the second model hatched better outcomes, highlighting the critical role these digital platforms play in shaping the behavior of a specific asset