Traditional asset pricing models assume rational investors who maximize expected utility based on all available information. However, real-world financial markets deviate from these idealized conditions due to investor heterogeneity and sentiment. This dissertation explores these deviations by addressing three key research questions. First, it examines the trading behavior of Robinhood investors around corporate announcements, comparing their impact on prices to other retail and institutional investors. Second, it investigates the role of investor sentiment—both retail and institutional—at the firm level and its influence on the cross-section of stock returns. Third, it assesses whether ChatGPT can be leveraged for stock selection, exploring the potential of artificial intelligence in financial decision-making.