Summary
The Rolling Average forecast algorithm is a very simple forecasting algorithm. It is present mostly for comparison purposes to methods commonly used for organizations coming off of spreadsheet-based approaches.
Rolling Average Forecast is used in our "benchmark" (6 month) forecast in the Forecast Error screen; to demonstrate how more sophisticated methods can provide forecast error benefits.
Settings
A rolling average is calculated over some target period, such as the previous 3 or previous 6 months of data. This is selectable by the user:
Average Length - This is the number of periods over which the rolling average is calculated
Calculation
So, a 3-month rolling average would be:
Forecast For Period X = (Actuals from Periods X-1 to X-3) / 3
In contrast to a simple spreadsheet method, In StockIQ, the simple rolling average algorithm is augmented by StockIQ's seasonality logic, so the rolling average can take advantage of any seasonality that we have detected. In this way, StockIQ's implementation of Rolling Average is more advanced than a typical rolling average.
However, the rolling average method is still quite limited in that it displays no awareness of trend, just an averaging of recent behavior.