Our study concerns retrospective learning, the characteristic feature of which is that prognostic uncertainty increases the more the further we look into the future. RL seeks to establish a metric for the outreach of prognostic scenarios. The purpose behind RL is to provide an easy-to-apply indicator, which informs non-experts about the time in the future at which a prognostic scenario ceases to be in accordance (for whatever reasons) with the system’s past. Ideally, this indicator should be derived concomitantly with building a prognostic model. RL concerns the limitations of predictions and prognostic scenarios.