Clearance
pricing.

This teaching model uses synthetic data and exact dynamic programming. The data and price schedules are illustrative, separate from Walmart’s production system. The published work combines deep reinforcement learning, simulation, and optimization.

Clearing stock takes priority over money in this model. With two weeks left, even the lowest price leaves stock unsold. Revenue after price-change costs excludes product and other operating costs.

Weekly demand is rounded from 8 × (100 ÷ price)^1.6. Prices can stay the same or decrease through $100, $80, $60, and $40. Sales cannot exceed remaining inventory. Each change, including the first drop from $100, incurs the selected cost while stock remains. The model omits demand uncertainty, competition, replenishment, and differences between stores.

Inspired by the multiobjective clearance problem in the paper coauthored by Viresh Jivane and the Walmart team. The goals are to clear inventory by a deadline, preserve revenue, and limit relabeling cost.