Track 12 · Time Series and Forecasting · Lesson 3 of 11
Moving averages
Compute trailing and centred moving averages, choose the window, and explain the delay, the lost end values and why a moving average is a poor forecast for a trending series.
30 minAdvanced7 graded exercisesMembers
Monthly orders bounce around so much that a manager cannot tell whether the business is growing. Averaging over a window of months irons out the noise and the seasons and shows the underlying direction — provided you know what the average hides and where it lags.
A moving average (MA) replaces each value by the mean of the values in a window around it. Two placements of the window matter:
- A trailing MA uses the current value and the before it. It can be computed up to the latest period, so it is usable in real time — but it describes the middle of its window, periods ago.
- A centred MA uses values on both sides. It describes the trend at that period without delay, but it needs future values, so it is missing for the most recent periods.
Included with a Formula Quest membership
It has 7 graded exercises with feedback, hints and worked solutions. After it you will be able to:
- Compute trailing and centred (including 2×m) moving averages
- Choose a window equal to the seasonal period to reveal the trend-cycle
- Explain the delay of a trailing MA and the missing recent values of a centred MA
- Explain why a moving-average forecast under-forecasts a trending series by about b(k+1)/2