Goal-based investing defines risk as the probability of an investor not attaining a goal. This definition captures both contextual information about the investor and distributional properties of investments that fall outside of the scope of Modern Portfolio Theory. Specifically, it can accommodate asymmetrical and “fat-tailed” distributions. In addition, it is a more intuitive definition of risk than volatility. This article introduces a forecasting method to leverage these two advantages. The proposed approach requires that humans create scenario-based forecasts. From these, computers derive a full probability distribution and, consequently, parameters such as expected return and volatility. These yield inputs for portfolio optimization. Lastly, forecasting skill can be measured directly using proper scoring metrics. Scenario-based forecasts are intuitive, scalable across investors, and transparently determine portfolio positions. Essentially, they attempt to leverage both the large amount of information embedded in human forecasts and the computationally efficient methods of academic finance and quantitative investing. On a more philosophical level, the human element removes the dependency on historical data and inductive logic that plagues quantitative modeling, thus avoiding the “Black Swan” problem.
Bill Wynne (Fri,) studied this question.