Joint Inventory Selection and Allocation Algorithms for JD Retail Supply Chain

In October 2024, JD Retail Supply Chain Technology won the Daniel H. Wagner Prize for its innovative inventory selection and allocation techniques. The paper introduces a data‑driven framework that integrates machine‑learning‑based selection algorithms (ML‑Top‑K and Reverse‑Exclude) with a novel end‑to‑end multi‑task learning allocation algorithm.

The two‑tier distribution network consists of Regional Distribution Centers (RDC) and Front‑line Distribution Centers (FDC). FDCs have limited capacity, so the problem is to decide which SKUs to stock in each FDC (selection) and how to transfer inventory from RDCs to FDCs (allocation) to maximize order fulfillment while minimizing stock‑out loss and transfer costs.

For the selection sub‑problem, the authors formulate an integer‑programming model and propose two heuristics: ML‑Top‑K, which predicts SKU sales using a hybrid TCN‑MLP architecture and selects the top‑K SKUs; and Reverse‑Exclude, which iteratively eliminates low‑impact SKUs until the capacity constraint is met. Both methods improve local order‑satisfaction rates compared with existing practices.

The allocation sub‑problem is modeled as a multi‑period stochastic inventory transfer problem. A new end‑to‑end learning algorithm combines demand forecasting, safety‑stock generation, and simulation‑based optimization in a unified framework. The model uses multi‑task learning, RNNs for temporal dependencies, and a simulation module to evaluate sales loss and transfer costs.

Extensive experiments on real JD data show that the selection algorithms increase local order‑satisfaction by 0.54pp (ML‑Top‑K) and 0.27pp (Reverse‑Exclude), and a combined approach yields a 2.2pp gain. The allocation algorithm achieves a 58.59% FDC satisfaction rate, surpassing baseline methods by over 1pp while keeping loss rates comparable.

Deployed across all eight RDCs and their associated FDCs, the system reduces inventory holding and capital costs by tens of millions of RMB annually, saves over a hundred million RMB in transfer costs, and improves inventory on‑hand rate by 0.85% and local order‑satisfaction by 2.19%.

These results demonstrate the practical impact of data‑driven optimization and AI techniques in large‑scale e‑commerce supply chain management.