<p>This research paper aims to leverage the computing power of Machine learning models to predict the stock prices for individual stocks and use them to optimize stock portfolios. The Extreme Gradient Boosting algorithm was used for prediction, and the dataset used was the Nifty 50 stock prices in the data range starting Jan 3rd, 2011 until July 31, 2024, stheced from Bloomberg Terminal. Technical features Average Directional Index and Stop and Reverse were added to the data, and then the data was normalized. The model was then trained on the normalized data. The average coefficient of determination obtained was 85%. The model accurately predicts expected returns. The future expected volatility can also be determined, which helps create highly optimized risk portfolios. The variability observed in the result values highlights the need for continued refinement and optimization of the model. Further analysis and hyperparameter optimization of the model can yield results with a higher level of accuracy. The implications of this research are significant for the finance industry, as it provides a robust algorithm for predicting expected returns and optimizing risk portfolios.</p>