Abstract

Within the framework of the importance (and costs) of reducing uncertainty, we present the argument that a good prediction isn’t about minimising statistical forecasting errors. In the process of formulating a prediction we break down uncertainty into three parts. Direction of change, timing of change and level of change are each able to assist us with day to day trading decisions. We show that the notion that it is necessary to precisely predict all three components to assist the decision making process is a fallacy. An analysis of the prediction making process of a number of Australian Fund managers is also provided.

Full Paper (Size: 355 KB)