In practice, when building models, one will use historical data to provide guidance on the value of the assumptions that should be used in the forecast. There are various ways that historical values could be used to information the forecast. For example, the forecast assumptions could:
- Be set to be exactly equal to the most recent figure historical figure.
- Use the average of historical figures (if several prior periods are available).
- Reflect trends in historical figures (for example, the growth rate over recent years may have been declining, so to use an average or a current figure could result in an overestimate).
- Use any historical information as a guide only, with forecast values reflecting not only the historical information (in some way), but also take into account other information that one is aware of (such as if a company were to shortly be introducing a major new product for the first time in several years).
A detailed discussion of these methods is beyond the scope here: At this point, we aim mainly to highlight that the transition point in the model (from historical to forecast) very often implicitly uses a “logic reversal”. The following provides an example in the context of growth-based assumptions, whilst the principles apply in many other situations.