Tracing money flows in cryptocurrency networks is notoriously difficult because funds rarely stay in one place. They move between accounts, across individual transactions, and through different token layers. A new paper on arXiv argues that this complexity demands a more expressive data structure than standard transaction graphs.

The authors propose a transaction-aware graph in which nodes represent account–asset states rather than just accounts or transactions. By making the graph explicitly aware of the transaction context, the model is designed to capture how value moves through multi-token systems in a single unified representation.

The abstract does not include experimental results or implementation details, so the paper's practical performance is not yet known. Still, the framing suggests a useful step toward more robust tools for analyzing cross-token flows, which could matter for compliance, forensics, and network research.