Static Pricing for Single Sample Multi-unit Prophet Inequalities

Computer scientists Pranav Nuti and Peter Westbrook have solved a fundamental algorithmic pricing problem in their paper 'Static Pricing for Single Sample Multi-unit Prophet Inequalities.' They address a scenario where a seller has k identical items to sell sequentially to buyers with unknown private values. The seller only has access to a single historical sample from the value distribution—a minimal data requirement common in new markets or for novel products. Their key breakthrough proves that for k≥3 items, setting a simple static price equal to the k-th largest observed sample value guarantees achieving at least 50% of the optimal possible social welfare (the total value created by transactions). This resolves a conjecture from prior work by Pashkovich and Sayutina that had only solved the cases for k=1 and k=2.

For markets with many items (large k), the researchers derived an even more powerful result. They proved the optimal static price using one sample is the (k-√(2k log k))-th largest sample. This strategy achieves a competitive ratio of 1 - √(2 log k / k), meaning it captures nearly all possible value as inventory grows—losing only a vanishing fraction. Crucially, they showed this is the best possible guarantee any static pricing scheme can achieve with just one sample, establishing a theoretical limit. The gap between this and the optimal ratio with full distribution knowledge (1 - √(log k / k)) is surprisingly small, demonstrating the power of simple, data-efficient algorithms.

This work bridges theoretical computer science and practical mechanism design, providing rigorously proven 'prophet inequalities'—bounds on how well online algorithms can perform compared to an omniscient prophet who knows the future. The single-sample model is particularly relevant for real-world platforms launching new services, where historical data is scarce. The results give marketplace designers confidence that simple threshold rules based on minimal data can yield strong performance guarantees, balancing revenue and welfare objectives in dynamic pricing environments.