It is argued that for specific forecast settings there must exist an optimally-sized model with forecasts that only need occasional adjustments by experts. The argument is built on recent evidence on the interaction between models and experts. A consequence of this is that the future research...