Researchers have developed a new method to identify centralized exchange (CEX) addresses on the Ethereum blockchain by analyzing their on-chain behavior. The study utilized Proof-of-Reserves disclosures from 15 CEXs and five non-CEX categories, deriving 58 behavioral features. A LightGBM model significantly outperformed a logistic regression model, achieving a macro PR-AUC of 0.9594 compared to 0.7697 with 17 selected features over a 730-day window. Key predictive signals identified through SHAP analysis included gas behavior, ERC-20 token activity, and internal transactions, offering a more interpretable approach to blockchain monitoring. AI
IMPACT Provides a more interpretable and accurate method for identifying cryptocurrency exchange addresses on the blockchain, potentially aiding in financial monitoring and security.
RANK_REASON Academic paper detailing a new classification framework and model performance. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →