Generative Adversarial Networks: A Systematic Review of Characteristics, Applications, and Challenges in Financial Data Generation and Market Modeling: 2019-2024

Abstract

Generative Adversarial Networks (GANs) have emerged as a promising solution for machine learning and artificial intelligence algorithms constrained by data availability and accessibility. Financial markets, alongside healthcare, present significant challenges due to data privacy and confidentiality concerns. GANs enable researchers to generate synthetic financial data that closely mirrors real-world datasets, facilitating advancements in market analysis and modeling. Despite their potential, a comprehensive evaluation of GAN-based financial data generation remains limited, necessitating a systematic assessment of existing methodologies and findings. This paper presents a systematic review of GAN architectures applied to financial data generation and market modeling. Our study is distinguished by its comprehensive exploration of various GAN variants and their specific applications within different facets of financial markets, including stock price prediction, algorithmic trading, portfolio optimization, risk management, and fraud detection. Leveraging thirty relevant papers from four major databases (IEEE Xplore, Web of Science, Scopus, and arXiv), we synthesized key findings, identify challenges, and highlight limitations in the application of GANs for financial data generation. Our findings reveal that while GANs enhance data privacy and accessibility, they also face limitations such as mode collapse, instability during training, and regulatory concerns in financial markets. This qualitative review provides valuable insights for researchers and stakeholders, offering a foundation for future studies and innovative applications of GANs in financial markets.