Application of ML and DL in Stock Price Prediction and Portfolio Optimization

This paper focuses on using ML and DL methods to predict stock prices and make portfolio recommendations. Utilizing a comprehensive dataset from reliable financial sources, the research employs four distinct models: Linear regression model, XGBoost algorithm, random forest, and LSTM. Feature engineering is employed in the study to pull out technical indicators to improve the model's performance. Portfolio management is carried out by using the Monte Carlo simulation approach. The research also builds a web application to showcase a firm's real-time stock price prediction. It is discovered that LSTM networks, for instance, predict better than the ML methods in the temporal characteristics of the stock price data. However, XGBoost and Random Forest are among the most accurate in prediction issues involving extensive feature extraction. From an ethical standpoint, the study emphasizes the importance of considering the implications of AI-based calculations of financial forecasts and the responsible use of such models in everyday financial markets. Possible drawbacks of the study, namely overfitting and constant changes in the financial markets, are considered. The paper concludes by underlining the need for a responsible and mindful approach to constructing a reliable, ethical model of the usage of AI in finance.