Financial news articles are a challenge to analyze because of the unique style and vocabulary utilized in the domain. Many terms are hyperspecific and may not be known to casual readers while being commonplace to an experienced reader, i.e. 'alpha-generating', 'floor', 'priced in', 'resistance', etc. Financial sentiment analysis is a niche field, plagued with many challenges. Most of the data is unlabeled, and general language models don’t work well on financial articles because of the unique style and vocabulary in said articles. My goal for this project is to find appropriate models to represent financial sentiment analysis and fine-tune them to work at a satisfactory level, and to find relationships and derive insights between articles and learn more about what makes the domain of financial news tick.