How to Compute a Rolling Average
To follow along with how to compute a rolling average, please download The Basic Rolling Average Forecast Example, as it will be used to explain the calculations in this section.
The first decision a company has to make when calculating a rolling average is how many periods will be averaged; known as "n." In the example, n = 4 periods. That is, four periods of historical data will be used to develop the rolling average. A company must choose the number of periods they want to average based on how reactive they want the rolling average to be with recorded data changes. The more periods averaged, the less reactive the rolling averages will be; which means using only a few periods, such as one or two, will provide very reactive rolling averages—but then, with that little data, you might as well just use a standard average.
Computing a rolling average requires data recorded over several consistent time periods. Usually, historical data, such as historical sales, production, or even profits made; is used. This rolling average produces a future value, known as a forecast. A forecast is a calculated prediction of any type of future data for the next business period; including daily, weekly, or monthly forecasts; based on the most recent number of periods, "n," of historically recorded data being used in the calculation.
More specifically, a rolling average can be defined as a continuously moving, calculated average of the most recent number of periods; "n," defined by the company. Let's take a look at the example to see how this calculation works.
In Table 1 of the example, the first forecast calculated is for period five; which is 775. This was calculated by averaging the four most recent historical pieces of data right before period five indicated with red check marks, since "n = 4 periods" for this example. The detailed calculations for period five's forecast are explained in Table 2. Once the actual data for period five is collected and recorded into the table, the forecast for period six can be calculated.
The rolling average forecast for period six is calculated based on the four most recent historical pieces of data before the sixth period; an average of the historical data for periods two through five, indicated with blue check marks. The forecast is then documented in the table, which is the blue "825" forecast for period six in Table 1 in the example. In order to see the detailed calculations for the sixth period's forecast, please see the second row of Table 2 in the example.
To find out how to compute a rolling average forecast using two variables, please continue reading on Page 2.