A Study on Methods, Challenges, and Future Directions of Generative Adversarial Networks in Stock Market Prediction
Mohammad Diqi, Ema Utami, Kusrini Kusrini, Ferry Wahyu Wibowo · 2024
Generative Adversarial Networks (GANs) have demonstrated a transformative capacity in financial modeling, particularly stock price prediction. This systematic literature review delves into the methodologies, challenges, and efficacy of GANs in this field, underscoring their ability to generate complex, high-dimensional data that traditional models struggle to process. By learning from the distribution of historical stock data, GANs can uncover intricate patterns missed by conventional approaches, offering a more nuanced understanding of market dynamics. However, their deployment is not without challenges, including training instability and the complex nature of financial data, which can lead to issues like mode collapse and non-convergence. The review aggregates insights from various studies to provide a comprehensive overview of current applications, highlight innovative methods, and outline future research directions to further harness GANs' potential in financial analytics. The ultimate aim is to illuminate how GANs could revolutionize traditional stock market analytics, transforming them into more predictive, robust tools for investors and analysts.