Thesis abstract:

«p»«p»Stock market prediction is a critical challenge in financial research due to the complexity of market behavior. This study explores the effectiveness of «strong»Extended Long Short-Term Memory (xLSTM)«/strong» models in forecasting stock prices by leveraging a combination of «strong»OHLCV data, technical indicators, and fundamental analysis«/strong». Unlike conventional models that rely solely on historical price data, this research integrates «strong»market trends, momentum indicators, and financial fundamentals«/strong» to enhance predictive accuracy.«/p» «p»The proposed xLSTM model is trained using «strong»deep learning techniques«/strong», incorporating optimized hyperparameters and feature selection strategies. Performance evaluation is conducted using key metrics, including «strong»Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R² score«/strong». Comparative analysis demonstrates that integrating «strong»technical and fundamental indicators«/strong» significantly improves stock price prediction accuracy over models relying only on OHLCV data.«/p» «p»The findings highlight the advantages of «strong»data-driven hybrid models«/strong» in financial forecasting, offering valuable insights for «strong»investors, traders, and financial analysts«/strong». This research contributes to the advancement of «strong»machine learning in finance«/strong», supporting algorithmic trading strategies and informed investment decisions in volatile markets.«/p» «h3» «/h3» «p» «/p» «p» «/p» «p» «/p»«/p»