Processes involving troposphere–stratosphere coupling have been identified as important contributors to an improved subseasonal to seasonal prediction in the mid-latitudes. However, atmosphere models still struggle to accurately predict stratospheric extreme events. Based on a novel approach in this study, we use ERA5 reanalysis data and ensemble simulations with the ICOsahedral Non-hydrostatic atmospheric model (ICON) to investigate tropospheric precursor patterns, localised troposphere–stratosphere coupling mechanisms, and the involved timescales of these processes in the Northern Hemisphere extended winter. We identify two precursor regions: mean sea level pressure in the Ural region is negatively correlated with the strength of the stratospheric polar vortex for the following 5–55 d with a maximum at 25–45 d, and the pressure in the extended Aleutian region is positively correlated with the strength of the stratospheric polar vortex the following 10–50 d with a maximum at 20–30 d. A simple precursor index based on the mean pressure difference of these two regions is very strongly linked to the strength of the stratospheric polar vortex in the following month. The pathways connecting these two regions to the strength of the stratospheric polar vortex, however, differ from one another. Whereas a vortex weakening can be connected to prior increased vertical planetary wave forcing due to high-pressure anomalies in the Ural region, the pathway for the extended Aleutian region is less straightforward. A low-pressure anomaly in this region can trigger a Pacific–North American-related (PNA-related) pattern, leading to geopotential anomalies of the opposite sign in the mid-troposphere over central North America. This positive geopotential anomaly travels upward and westward in time, directly penetrating into the stratosphere and thereby strengthening the stratospheric Aleutian High, a pattern linked to the displacement towards Eurasia and subsequent weakening of the stratospheric polar vortex. Overall, this study emphasises the importance of the time-resolved and zonally resolved picture for an in-depth understanding of troposphere–stratosphere coupling mechanisms. Additionally, it demonstrates that these coupling mechanisms are realistically reproduced by the global atmosphere model ICON.