Portfolio Optimization Using a Hybrid BI-LSTM Model for NIFTY 50

Traditional portfolio optimization models using technical or fundamental analysis do not react to shifting market conditions. Technical analysis relies on historical price trends, which are likely to miss short-term market movement, while fundamental analysis relies on long-term figures but is likely to miss key market movement. This study suggests a hybrid model integrating Bi-LSTM-based stock return forecasting with significant financial metrics such as P/E ratio, Piotroski score, and dividend yield. Using a 10-year Nifty 50 data set, the model's Sharpe Ratio is 8.86, expected return is 30.44%, and portfolio risk is 2.8%. It outperforms traditional and machine learning models by optimizing risk-adjusted returns while maintaining diversification. The approach gives a solid solution to optimize the investment strategy with equilibrium among short-run responsiveness and long-run financial health.