Epistora Journal of Artificial Intelligence and Intelligent Systems
Current Issue

Artificial intelligence and machine learning have increasingly transformed financial decision-making, yet many retail investors continue to rely on conventional valuation ratios for stock selection. This study examines whether machine learning models can provide a practical and accessible framework for screening Indian equities using publicly available National Stock Exchange (NSE) data. A Random Forest classifier was developed using six fundamental and momentum-based financial features and evaluated against traditional P/E and P/B ratio screening methods. The results demonstrate that the machine learning approach achieved superior precision, stronger portfolio performance, and improved risk-adjusted returns, with the greatest advantages observed in under-researched mid-cap Information Technology and Pharmaceutical sectors. Feature importance analysis revealed that momentum and quality indicators contributed more significantly than conventional valuation metrics. While the study confirms the potential of machine learning as a practical screening tool for retail investors, it also recognizes limitations arising from survivorship bias, limited sample size, imperfect data quality, and a single-year validation period. Future research should expand the analysis across the full Nifty 500 universe, incorporate natural language processing of corporate disclosures, and investigate reinforcement learning for dynamic portfolio rebalancing. Overall, the findings suggest that freely available data and open-source machine learning tools can meaningfully improve equity screening while remaining accessible to individual investors.