Shadow trading detection: a graph-based surveillance approach

This paper introduces a novel graph-based deep learning framework for detecting risks of shadow trading, an emerging form of insider trading where material nonpublic information is used to trade securities of economically related but distinct companies. Motivated by the landmark SEC v. Panuwat case in April 2024, the study proposes an Adaptive Market Graph Intelligence Network (AMGIN) that integrates both industry relationships (e.g., sectoral ties, inter-organizational connections) and dynamic market behaviors (e.g., short/long-term price co-movements) to uncover hidden trading patterns. By modeling the financial market as a spatio-temporal graph, the framework captures complex interdependencies that traditional statistical methods often overlook. Empirical evaluation using US equity market data demonstrates AMGIN’s superior ability to identify subtle, non-obvious relationships indicative of shadow trading, offering regulators a scalable, data-driven tool for modern market surveillance.