Financial-Market Forecasting and Modelling from Econometrics to AI: An Integrated Systematic and Bibliometric Review with Content Synthesis (1990–2024)

This study offers a comprehensive assessment of financial market modeling through a PRISMA-based systematic review, bibliometric analysis, and content synthesis. We examined 67 review articles (1990–2024) from Web of Science to build a conceptual framework, and 4982 articles (1990–2024) were analyzed with Biblioshiny. Five main clusters emerge: AI and deep learning for prediction; hybrid models that combine traditional and computational approaches; theoretical foundations, including the Efficient Market Hypothesis and critiques; high-frequency prediction and volatility analysis; and modeling of cryptocurrencies and digital assets. Temporal patterns show a shift from traditional econometrics to hybrid and deep learning methods, heightened attention to uncertainty and volatility during crises, rapid growth in crypto-focused modeling, and increased use of sentiment/news data after 2017. The content analysis highlights key gaps and future directions: standardized open benchmarks and reproducible frameworks; regime-sensitive validation; interpretable hybrid models that merge econometric structure with machine-learning flexibility; and wider applicability across assets, markets, and data types. The study provides a structured guide to intellectual and applied modeling, supporting future advances in forecasting, risk management, and policy design.