Do Better Volatility Forecasts Lead to Better Portfolios? Evidence from Graph Neural Networks

TL;DR

This study investigates whether graph neural networks enhance volatility forecasts and if such improvements translate into better portfolio performance, revealing that different models excel in different objectives.

Contribution

It demonstrates that improved volatility forecasting does not necessarily lead to better portfolios, emphasizing the importance of model choice based on specific financial objectives.

Findings

Models with lowest forecast MSE, highest ranking accuracy, and best Sharpe ratio are different.

Forecast accuracy, ranking, and portfolio performance are related but distinct objectives.

Graph models add value only when the portfolio rule exploits the encoded structure.

Abstract

This paper tests whether graph neural networks improve realized volatility forecasts and whether those forecasts improve portfolio performance. Using weekly realized volatility for 465 S&P 500 equities from 2015-2025, Heterogeneous Autoregressive and Long Short-Term Memory baselines are compared against GraphSAGE models built on rolling correlation, sector, and Granger-causal graphs, with and without macro regime features. The empirical finding is that the model with the lowest forecast MSE, the model with the highest cross-sectional ranking accuracy, and the model with the highest portfolio Sharpe ratio are three different models. Forecast accuracy, ranking quality, and portfolio performance are related but not interchangeable objectives. Graph volatility models add value only when the portfolio rule can exploit the cross-sectional structure they encode.