Researchers introduce MELT, the first behavioral trace dataset designed to analyze and detect high-risk memecoin launches on Solana. The dataset parses massive blockchain activity into typed behavioral records, revealing coordination strategies and ownership concealment among coordinators
Launchpads pose significant risk to investors in the memecoin space, necessitating advanced detection methods beyond traditional methods. To address this, the researchers introduced MELT (MEmecoin Launch Trace), a novel dataset comprising over 41,000 memecoin launches and 200 million transactions parsed into detailed behavioral records (swaps, wash trades, transfers, and mints). MELT uniquely provides bundle-trace data linking coordinated accounts, demonstrating that an average of 36.5% of token supply is held by coordinated entities—a strategy designed to conceal true ownership concentration.
The dataset includes 122 behavioral features and risk-level annotations, enabling supervised learning models to detect high-risk launches at a population scale. Benchmarking ML models on this data shows that integrating these behavioral traces can significantly reduce investment loss, translating behavioral data directly into actionable risk mitigation strategies.