Trading Graph Neural Network
This addresses the need for more accurate price prediction in trading networks, though it appears incremental as it hybridizes existing techniques.
The paper tackles the problem of estimating how asset, dealer, and relationship features affect prices in trading networks by proposing Trading Graph Neural Network (TGNN), which combines simulated method of moments with graph neural networks and outperforms existing reduced-form methods in prediction accuracy.
This paper proposes a new algorithm -- Trading Graph Neural Network (TGNN) that can structurally estimate the impact of asset features, dealer features and relationship features on asset prices in trading networks. It combines the strength of the traditional simulated method of moments (SMM) and recent machine learning techniques -- Graph Neural Network (GNN). It outperforms existing reduced-form methods with network centrality measures in prediction accuracy. The method can be used on networks with any structure, allowing for heterogeneity among both traders and assets.