We use a dedicated simulation setup to shed light on the question whether, and why, some sudden stratospheric warming events are more likely than others to develop a surface response. We find that the propensity for downward coupling is unique to each event, and that from day one, the chance of a lower-stratospheric response can be predicted – a key step toward anticipating the surface response, moving beyond 'random' surface outcomes to quantified likelihoods of the ensuing surface response.