Thompson Sampling

Thompson sampling is a Bayesian approach to the exploration-exploitation problem that maintains posterior probability distributions over each arm's reward parameters and makes decisions by sampling from these distributions, selecting the arm with the highest sampled value. This elegant probabilistic method naturally balances exploration and exploitation: arms with uncertain rewards remain viable candidates, while arms with proven high performance get selected more frequently. Thompson sampling is theoretically sound, computationally efficient, and empirically superior to many alternatives, making it a preferred approach in practical applications requiring adaptive decision-making. Example: Yearn Finance's automated yield farming strategies could employ Thompson sampling to adaptively allocate capital across competing yield strategies, maintaining uncertainty estimates over each strategy's returns and sampling to balance trying new opportunities with exploiting known performers. Why it matters for AI and data in Web3: Thompson sampling provides a principled, Bayesian framework for Web3 protocols and DeFi platforms to automatically allocate capital and adjust strategies across opportunities, combining exploration of emerging opportunities with exploitation of proven performers while maintaining mathematical rigor and avoiding over-commitment to suboptimal choices.

Explore the full Web3 Glossary — 2,062+ expert-curated definitions. Need guidance? Talk to our consultants.