In an era where financial markets are increasingly influenced by the rapid dissemination of information through various media, understanding market sentiment has become crucial for investors and financial analysts. Sentiment analysis, a field at the intersection of finance and natural language processing (NLP), offers valuable insights into market trends and investor behaviour. Recognizing the potential of advanced NLP in the finance domain, this study presents an adaptation of BERT for sentiment analysis within the stock market domain, achieving an overall accuracy of 77 percent. Through meticulous preprocessing—including tokenization, lemmatization, and standardization of text—combined with the advanced capabilities of BERT's pre-trained models, the research aimed to capture and classify market sentiment from textual data accurately. The fine-tuned model demonstrated a precision of 72 percent for negative sentiment detection and 80 percent for positive, with recall rates at 61 percent and 87 percent, respectively. The resulting F1-scores were 66 percent for negative and a robust 83 percent for positive sentiments, indicating a more reliable identification of positive over negative sentiment. These metrics affirm the model's effectiveness in discerning sentiment polarity and point to potential areas for enhancement, particularly in the Recall of negative sentiment. The findings underscore the significant promise of employing sophisticated NLP methodologies like BERT in financial sentiment analysis, which could be transformative for fields such as algorithmic trading and economic prediction by leveraging the subtle nuances captured through sentiment classification.