Nonparametric Contextual Online Bilateral Trade

Researchers developed a novel algorithm for contextual online bilateral trade that operates under extreme constraints. It uses hierarchical trees to handle arbitrary nonparametric valuation functions with only one-bit feedback (trade/no-trade) while maintaining strong budget balance. The algorithm achieves a tight regret bound of Õ(T^{(d-1)/d}), matching a new lower bound. This represents a significant advance over prior linear models, enabling smarter automated pricing in complex, real-world market environments.