SpotV2Net: Multivariate Intraday Spot Volatility Forecasting via Vol-of-Vol-Informed Graph Attention Networks

This paper introduces SpotV2Net, a multivariate intraday spot volatility forecasting model based on a Graph Attention Network architecture. SpotV2Net represents financial assets as nodes within a graph and includes non-parametric high-frequency Fourier estimates of the spot volatility and co-volatility as node features. Further, it incorporates Fourier estimates of the spot volatility of volatility and co-volatility of volatility as features for node edges. We test the forecasting accuracy of SpotV2Net in an extensive empirical exercise, conducted with high-frequency prices of the components of the Dow Jones Industrial Average index. The results we obtain suggest that SpotV2Net shows improved accuracy, compared to alternative econometric and machine-learning-based models. Further, our results show that SpotV2Net maintains accuracy when performing intraday multi-step forecasts. To interpret the forecasts produced by SpotV2Net, we employ GNNExplainer, a model-agnostic interpretability tool and thereby uncover subgraphs that are critical to a node's predictions.