Mastering Moving Average Methods: Simple, Weighted, and Trend Techniques

1 Moving Average Method

The moving average method calculates sequential averages of a specified number of observations in a time series to reflect long‑term trends, smoothing out periodic and irregular fluctuations. It includes simple moving average, weighted moving average, and trend moving average techniques.

2 Simple Moving Average Method

When the target's basic trend fluctuates around a level, the Simple Moving Average (SMA) can be used to build a forecast model. It averages recent observations and is suitable for short‑term forecasts when the trend does not change significantly. The SMA predicts future periods by taking the average of the most recent values; the window size should be larger when random variation dominates and smaller when the trend changes rapidly. For data with a known seasonal cycle, the window should equal the cycle length, and the optimal window can be chosen by comparing forecast errors.

3 Weighted Moving Average Method

In the simple moving average, each observation contributes equally, but recent data often contain more information about the future. The Weighted Moving Average (WMA) assigns larger weights to recent observations and smaller weights to older ones. The forecast value for period t is computed as the weighted sum of past observations divided by the sum of weights. The specific weights are chosen based on experience, giving higher importance to recent data.

4 Trend Moving Average Method

This approach uses the lag of moving averages to build a linear trend forecast model. Assuming a time series follows a straight‑line trend from a certain point onward, the model can be expressed as Y = a + b·k, where k is the period index, a is the intercept, and b is the slope (smoothing coefficient). The smoothing coefficient is derived from the moving average values, allowing the method to capture both linear trends and periodic fluctuations.

Reference

ThomsonRen GitHub https://github.com/ThomsonRen/mathmodels