Data sets resources.

The step-up effect of great coaches on Olympic performance has been widely recognized, but its dynamic influence mechanism and quantitative evaluation remain methodological challenges. This study proposes a Change-Point Driven Difference-in-Differences with Decay Model (), which integrates the CUSUM algorithm, dynamic difference method (DID), and exponential decay function. The Great Coach effect’s nonlinear characteristics and time attenuation rule were analyzed systematically. First, based on the improved CUSUM algorithm to detect the abrupt points of performance, the dual test mechanism of medal continuity and competition size stability was introduced to filter the pseudo-abrupt signals (such as the host effect and short-term strategic interference). Secondly, a hierarchical DID model was used to quantify the net effect of heterogeneous coach turnover events to solve the problem of traditional methods ignoring the dynamic confounding bias and run-in period. Finally, the sustainability difference of the coaching effects is revealed by the half-life model. Empirical studies show that the effect half-life of a systematic coaching system (such as Zhou Jihong coaching the Chinese diving team) is more than 20 years, while the technology-driven intervention (such as the AI tactical optimization for Japanese judo) has a half-life of only 5.3 years. The model predicts that in the 2028 Los Angeles Olympics, the introduction of great coaches will enable the Brazilian swimming team to achieve a breakthrough from 0 to 5 medals (95% CI: 4.7–6.3). This study provides an explainable and predictive framework for the allocation of coaching resources in Olympic strategy, and its methodology can be extended to dynamic causal inference in policy evaluation and organizational management.