AI-Based Stock Return Predictions in Nascent and Emerging Economies: A Comparative Analysis Using Machine Learning Models

Abstract

Financial markets have an important role in encouraging economic development by handling resources and providing investment venues that inspire growth. However, standard statistical and econometric models often emphasize the non-linear and volatile dynamics of stock returns, specifically in developing countries. This study investigates the predictive power of Artificial Intelligence (AI)-based models in forecasting stock returns across nascent and emerging economies through a comparative meta-analytical approach. Using machine learning models, namely Random Tree and Multilayer Perceptron (MLP) models, the research estimates the predictive accuracy of market maturity, data availability, and institutional structure. The results show that due to enhanced data integrity, systematized return patterns, and augmented market stability, AI models display improved prediction accuracy and dependability in rising markets, such as China, India, and Saudi Arabia. Conversely, emerging markets with weaker predictive frameworks and greater volatility, such as Nigeria, Vietnam, and Pakistan, result in poorer model performance. The support for both hypotheses (H1 and H2) shows that market maturity greatly intensifies the efficacy of AI models in financial forecasting. By highlighting the dynamic role which institutional growth and data robustness contributes towards enhancing stock return prediction, this study complements the expanding body of research on AI-driven financial analytics. The findings highlight the potential of AI to improve forecasting accuracy, reduce market inefficiencies, and support financial stability in emerging economies, offering insightful information to investors, politicians, and financial technologists.