Global Neural Networks and The Data Scaling Effect in Financial Time Series Forecasting

2023-09-05

AI Summary

This paper addresses the poor generalizability and unstable empirical performance of local modeling for financial time series—particularly volatility—under data scarcity. We propose a global neural network training paradigm that pools cross-sectional data from tens of thousands of global equities and employs end-to-end deep learning to jointly model large-scale heterogeneous time series. We systematically uncover, for the first time, a positive scaling law between data scale and heterogeneity in financial forecasting. The model achieves robust out-of-sample volatility prediction for unseen individual stocks and portfolios using only 12 months of training data. It exhibits adaptive responsiveness to structural breaks and robustness to outliers, while faithfully reproducing canonical stylized facts—including volatility clustering and long memory—in an interpretable manner. Empirical results demonstrate substantial improvements in both out-of-sample prediction accuracy and generalization capability across diverse market regimes.

Abstract

Neural networks have revolutionized many empirical fields, yet their application to financial time series forecasting remains controversial. In this study, we demonstrate that the conventional practice of estimating models locally in data-scarce environments may underlie the mixed empirical performance observed in prior work. By focusing on volatility forecasting, we employ a dataset comprising over 10,000 global stocks and implement a global estimation strategy that pools information across cross-sections. Our econometric analysis reveals that forecasting accuracy improves markedly as the training dataset becomes larger and more heterogeneous. Notably, even with as little as 12 months of data, globally trained networks deliver robust predictions for individual stocks and portfolios that are not even in the training dataset. Furthermore, our interpretation of the model dynamics shows that these networks not only capture key stylized facts of volatility but also exhibit resilience to outliers and rapid adaptation to market regime changes. These findings underscore the importance of leveraging extensive and diverse datasets in financial forecasting and advocate for a shift from traditional local training approaches to integrated global estimation methods.

Problem

• Improves financial time series forecasting accuracy
• Leverages global datasets for robust predictions
• Adapts to market changes and volatility

Innovation

• global neural networks
• data scaling effect
• volatility forecasting improvement