Predicting stock prices and market direction is a long-standing problem in finance and computer science, and artificial intelligence (AI) is widely proposed as a way to solve it. This paper presents a comparative case study of six recent publications on AI-based stock market prediction, covering classical machine learning, deep learning, metaheuristic hyperparameter optimization, and sentiment analysis. After reviewing the market-theory background and the standard prediction pipeline, we re-examine the reported results of four empirical studies: linear regression, LSTM, and nine classifiers on Apple stock and tweet sentiment; seven classifiers on seven G7 stock indices; ANN and CNN models on India’s NIFTY index; and genetic-algorithm and artificial-rabbits-optimized LSTMs on the 30 Dow Jones stocks. Summary statistics recomputed from the published tables show that (i) the best algorithm depends on the market, with ANN best on four of seven G7 indices and logistic regression on the other three; (ii) metaheuristictuned LSTMs cut mean MAPE from 3.9–4.5% to about 2.0% relative to untuned networks, yet the two optimizers differ by only 0.04 percentage points; and (iii) headline accuracies ranging from roughly 50% to nearly 99% are not comparable, because the studies use different targets and metrics. We conclude that AI methods capture useful short-horizon structure, but none of the reviewed evidence demonstrates reliable, cost-adjusted outperformance of the market.

Anurag Ajay Satam, Dr. Manoj Singh, "Artificial Intelligence in Stock Market Prediction: A Comparative Case Study of Machine Learning and Deep Learning Approaches" International Journal of Scientific Research and Engineering Development, V9(5): Page(1131-1138) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.