 This paper proposes a framework to detect anomalies in peer-to-peer transactions on a specific NFT marketplace. The authors first built a model to estimate the profit gain from selling a particular collectible on the platform. They then used random forest conditional density estimation, RFCDE, to identify transactions which deviate from the expected profits. Finally, they analyzed the trade networks formed from these anomalous transactions and compared them with the full trade network of the platform. Their results showed that the two networks were statistically different in terms of edge density, closure, node centrality, and node degree distribution. While this does not necessarily imply that the transactions are illegal, they should be further investigated by relevant authorities to determine if they are indeed illicit. This article was authored by Konstantinos Pelikrini's, XIN Liu, Prashant Krishnamurti, and others.