Analysis of Twitter Sentiment to Predict Financial Trends

This paper attempts to predict financial trends by analyzing tweets on the Twitter micro-blogging platform. It provides an enhanced model that builds on the functionality of the pre-existing RoBERTa large model that can be used to determine the general sentiment of financial tweets. The model is fine-tuned to work best with financial tweets, with an ideal hyperparameter configuration given for an easy-to-train, accurate model. A case study is also provided regarding one potential use of this model by utilizing a Twitter API to access and analyze the latest tweets for financial keywords, and providing an interface for users to be able to access and understand the sentiment and trends surrounding the keyword.