MarketGANs: Multivariate financial time-series data augmentation using generative adversarial networks

2026-01-25

AI Summary

This study addresses the challenge of modeling cross-asset dependencies and extreme-event co-movements in high-dimensional financial return data, where scarcity of observations hinders reliable estimation. To this end, the authors propose MarketGAN, a generative adversarial network that incorporates asset pricing factor structures as economic inductive biases. MarketGAN uniquely embeds an explicit factor model within the GAN framework and leverages temporal convolutional networks (TCNs) to dynamically capture time-varying factor loadings and volatilities. This approach effectively reproduces complex statistical features of financial markets, including cross-sectional dependence, tail co-movements, volatility clustering, and long-range temporal dependencies. Empirical experiments on daily U.S. equity data demonstrate that the generated samples closely replicate real-market statistical properties, and covariance matrices estimated from synthetic data yield significantly improved portfolio optimization performance compared to conventional methods, delivering clear economic value.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

This paper introduces MarketGAN, a factor-based generative framework for high-dimensional asset return generation under severe data scarcity. We embed an explicit asset-pricing factor structure as an economic inductive bias and generate returns as a single joint vector, thereby preserving cross-sectional dependence and tail co-movement alongside inter-temporal dynamics. MarketGAN employs generative adversarial learning with a temporal convolutional network (TCN) backbone, which models stochastic, time-varying factor loadings and volatilities and captures long-range temporal dependence. Using daily returns of large U.S. equities, we find that MarketGAN more closely matches empirical stylized facts of asset returns, including heavy-tailed marginal distributions, volatility clustering, leverage effects, and, most notably, high-dimensional cross-sectional correlation structures and tail co-movement across assets, than conventional factor-model-based bootstrap approaches. In portfolio applications, covariance estimates derived from MarketGAN-generated samples outperform those derived from other methods when factor information is at least weakly informative, demonstrating tangible economic value.

Problem

• financial time-series
• data scarcity
• cross-sectional dependence
• tail co-movement
• asset returns

Innovation

• MarketGAN
• generative adversarial networks
• factor models
• temporal convolutional network
• financial time-series augmentation