Decathlon implements Chronos-2 for large-scale demand forecasting

Decathlon's Forecasting Challenge

Decathlon, a large sporting goods retailer with over 100,000 employees and 400 million users, requires accurate demand forecasting to ensure product availability. The company faces challenges due to its scale and product diversity, which includes tens of thousands of products across more than 80 sports with seasonal demand patterns.

The forecasting system predicts weekly sales for all products across two horizons: a 12-week replenishment window for purchase planners and a 52-week strategic horizon for long-term stock and capacity planning. This system runs weekly across multiple supply zones, including Europe, India, China, South East Asia, and Latin America, with each zone covering up to 25,000 products.

Transition to Chronos-2

Decathlon selected Chronos-2 as a core component of its forecasting stack after evaluating multiple time series foundation models. This move follows previous approaches that included a hybrid system from 2021-2024 using Amazon SageMaker AI DeepAR for short-term forecasts and Holt-Winters exponential smoothing for longer terms. DeepAR was retrained weekly to adapt to recent trend shifts.

In 2024, Decathlon introduced Temporal Fusion Transformer (TFT) with covariates, which offered improved accuracy for long-horizon forecasts. The transition to Chronos-2 addresses the operational overhead associated with these prior methods.