Portfolio optimization with quantitative and newspaper sentiment analysis

This thesis explores integrating sentiment analysis of newspaper articles into portfolio selection based on quantitative analysis. The empirical study proposes an optimization model incorporating both analyses to achieve performance enhancement over selected benchmark. By incorporating sentiment-driven adjustments into traditional quantitative measures such as the Historical volatility, Sharpe Ratio, and Jensen’s Alpha, the optimized portfolio achieves better returns over a period of 5 years. Through an extensive analysis, it is established that sentiment analysis positively impacts the model and shows great potential in actively managed portfolios. The model was tested under different scenarios, where one included the application of a 1~\% transaction charge and reduced trade volumes. The results indicate that although the sentiment-enhanced model shows competitive performance against the NASDAQ Composite index, certain factors, such as significant investments in assets with unpredictable price jumps, highlight the need for further refinements. These improvements are left for future extension of this work. Overall, the findings underscore the potential of sentiment analysis in financial modeling while also pointing out areas for future research and optimization.