AI- Driven On-Chain Behavioral Pattern Discovery for Whale Sentiment in US Crypto Markets

Abstract

This study investigates the use of artificial intelligence to uncover behavioral patterns of whale participants in US cryptocurrency markets and to infer their impact on market sentiment. By leveraging on-chain transaction data and exchange activity, the research constructs behavioral features capturing transaction frequency, transfer volume, and wallet clustering. Machine learning models, including tree-based learners, recurrent neural networks, attention-enhanced architectures, and ensemble frameworks, are employed to identify distinct whale archetypes and to translate their activity into predictive sentiment signals. The findings reveal differentiated behavioral strategies among whales, with high-frequency accumulators, occasional large movers, and directional distributors exerting unique influences on market dynamics. Furthermore, attention mechanisms and feature importance analysis enhance the interpretability of model predictions, enabling insights into the timing and nature of whale-driven market movements. Overall, the study demonstrates the potential of AI-driven on-chain analytics to provide actionable intelligence for traders, exchanges, and regulators, bridging the gap between raw blockchain data and meaningful behavioral insights.