The Silk Road pattern (SRP) is a leading mode of Eurasian atmospheric variability in boreal summer. It remains challenging for many models to predict the temporal phase of the SRP. This study investigates whether the forcing mechanism of the SRP can explain SRP prediction difficulty. Rossby‐wave sources (RWS) associated with the SRP are first identified in a reanalysis. An idealised barotropic model is then used to test wave propagation from all identified RWS, to isolate the relevant forcing locations of the SRP, namely the RWS hotspots. In addition to previous findings, a new hotspot in the central North Pacific is located, which can force the SRP by westward dispersion of a zonally elongated Rossby wave. Furthermore, a new mechanism of the Indian summer monsoon in forcing the SRP is discovered. Results also suggest that Rossby‐wave propagation by itself only explains a small fraction of the SRP amplitude, consistent with baroclinic energy conversion from the mean state, which has previously been shown to be vital for SRP growth—not represented in the idealised model. Seasonal hindcasts from the Pusan National University (PNU) coupled general circulation model are also analysed. The hindcasts can predict the SRP spatially but not temporally. Within a seasonal timescale, the hindcast develops an upper‐level mean‐state wind bias compared with the reanalysis, which is shown by the barotropic model to affect RWS hotspot locations of the SRP. While the reanalysis SRP is associated with North Atlantic sea‐surface temperature (SST) anomalies, the hindcast SRP is associated with tropical Pacific SST. Wind biases in the North Atlantic jet exit, subtropical western Pacific, and tropical Pacific are found to alter wave propagation from the North Atlantic and the North and tropical Pacific, respectively, and can explain the association of the SRP with different SSTs in the reanalysis and hindcast. This sensitivity of hotspots to mean‐state winds is proposed to reduce SRP prediction skill.