ABSTRACT

Equations....................................................................................................................................... 661 References ..................................................................................................................................... 664

There may be no statistical practice more mysterious than forecasting. One reason for this mystery is that the word ‘‘forecasting’’ is attached to many different practices. These practices involve discovering something unknown, which is, at least in principle, in the future. From a statistical point of view, forecasting breaks a cardinal rule: ‘‘do not generalize outside the range of the data.’’ The future is always outside the range of the data, so, forecasters are statistical renegades.