In this a paper a non-linear macro-stress testing methodology with focus on early warning is developed. The methodology builds on a variant of Random Forests and its proximity measures. It develops a framework in which naturally defined contagion and feedback effects transfer the impact of stressing a relatively small part of the observations on the whole dataset and thus allow to estimate the a stressed future state. It will be shown that contagion can be directly derived from the proximities while iterating the proximity based contagion leads to nat- urally defined feedback effects. This procedure allows accurate forecast of events under stress and thus allows to forecast the emergence of a potential crisis. The framework also estimates a set of the most influential economic indicators lead- ing to the potential crisis, which can then be used as indications of remediation or prevention of that crisis. Since the methodology is Random Forests based the framework is suitable for big data analysis.