Detecting pump-and-dump schemes involving cryptoassets with high-frequency data is challenging due to imbalanced datasets and the early occurrence of unusual trading volumes. To address these issues, we propose constructing synthetic balanced datasets using resampling methods and flagging a...