A Systematic Review of Stock Market Forecasting Utilizing Machine Learning and Deep Learning

This comprehensive review examines advancements in stock market trend prediction using advanced computational techniques, focusing on deep learning methods, such as Long Short-Term Memory (LSTM) and technical analysis. From an initial pool of 5, 1 0 4 scholarly articles, 5 5 studies were rigorously selected based on predefined inclusion and exclusion criteria for in-depth analysis. LSTM appeared in approximately 40% of these studies, reflecting their capability in managing sequential time-series data and mitigating the vanishing gradient problem. The review also highlights a growing interest in Although less frequently employed, Transformer models possess significant potential for handling large datasets and modeling long-term dependencies. Many studies combined these advanced models with technical indicators, including Moving Averages and Relative Strength Index, to improve predictive accuracy. However, only 35.3% of the studies evaluated the profitability of these forecasting models, revealing a gap between predictive accuracy and real-world financial performance. The review advocates for a holistic approach encompassing diverse datasets, advanced trading strategies, and a heightened emphasis on profitability to better synchronize model outcomes with tangible financial objectives. © 2025 IEEE.