One of the significant problems with scenario generation is the sentiment that scenarios are simply an individual’s or group’s guess as to future states of the world. This makes scenario-based risk management a guessing game, albeit one that might carry the weight of experts. Another issue with scenarios, especially in applications to stress testing of portfolios, is that one cannot know whether a scenario stresses a portfolio until after the fact. In this paper we show how to systematize the generation of scenarios so as to enable them to be generated completely automatically, without any prior assumptions on the underlying probability distributions. The only inputs required are the future macro events, financial or otherwise, that trigger the request for forward looking scenario analysis. Our methodology has one other important feature in that we can show that in a significant number of cases it guarantees the generation of spanning set of scenarios. That is, the worst and best scenarios are guaranteed to be in the generated set. More importantly, it minimizes the bias introduced when humans design scenarios. We conclude by showing a detailed example of a proof of concept of our algorithmic approach to scenario generation in the context of power distribution planning in California.