The financial market is undergoing a significant transformation with the advent of Generative AI, marking a paradigm shift in its design. This groundbreaking technology leverages advanced intelligent systems and extensive datasets to reshape the financial domain. In this study, we present SentEM-GEN, an advanced Generative AI model designed to identify and explain user sentiments and emotions. SentEM-GEN follows a text-to-text generative framework, incorporating dual selective attention mechanisms to optimize sentiment and emotion analysis. The model consists of two variants: Variant-1, which focuses on sentiment analysis, and Variant-2, which emphasizes emotional aspects and includes BART-Large for nuanced causal generation. We trained SentEM-GEN using the FinEMA dataset, a carefully constructed dataset annotated with sentiment, emotion, and cause labels. FinEMA analyzes shifts in the financial stock market based on various sources, such as tweets, financial reports, and news articles, achieving 93% accuracy in sentiment classification and 81% in emotion detection—surpassing traditional machine learning models. To demonstrate the versatility of our model and dataset, we conducted some extensive comparisons. Our findings suggest a positive trend in the stock during the first half of 2024, strengthening investor confidence despite existing uncertainties. This research highlights the potential of Generative AI to enhance our understanding of public sentiment and its influence on stocks, assets, mutual funds, bonds, and investments, offering valuable insights into the evolving landscape of finance. Code: https://github.com/ankan8145/SentEM_GEN