Exponential Moving Averages

Exponential Moving Averages, similar to Weighted Moving Averages, also assign a greater weight to more recent data values. Unlike Weighted Moving Averages, however, they use the previously calculated Exponential Moving Average value as a basis for calculation rather than the original (non-Averaged) data values. In this way, the calculation method used by Exponential Moving Averages is cumulative, meaning that (unlike Simple Moving Averages or Weighted Moving Averages) all previous data values have some effect on the Exponential Moving Average to be calculated, although this effect diminishes greatly with time.

Exponential Moving Averages tend to be more accurate than the other types of Moving Average when the original data values show a more rapid degree of variation over time (or other variable).

The formula for calculating an Exponential Moving Average (EMA) is:

X = (K * (C - P)) + P

Where:

X = Current EMA (i.e. EMA to be calculated)

C = Current original data value

K = Smoothing Constant

P = Previous EMA

(The first EMA in the range to be calculated is arbitrary and can be the corresponding original data value or, often, a Simple Moving Average value.

Where:

K = Smoothing Constant = 2 / (1 + n)

n = number of periods for EMA i.e. the Window to calculate.

This rather complex calculation is, perhaps, best illustrated by example.

Consider the table of monthly Sales values as shown previously:

Image shows Sales values.

If we calculated the Exponential Moving Average in a similar fashion to the 3-Month Simple Moving Average, we would perform the following steps:

  To calculate the exponential moving average:

  1. Calculate the Smoothing Constant according to the 2 / (1 + n) formula.

    N = window of values = 3, therefore the Smoothing Constant is:

    2 / (1 + 3) = 0.5
  2. For the first Exponential Moving Average, use the first original data value (in this case, that for the Month of "Jan").

  3. For subsequent values, follow the calculation according to the above formula.

    X = (K * (C - P)) + P as indicated below:

    Table 178. X = (K * (C - P)) + P Example

    Month

    Sales

    Calculation (K * (C - P)) + P

    Exponential Moving Average

    Jan

    10

    = Original value

    10

    Feb

    15

    (0.5 * (15 - 10)) + 10

    12.5

    Mar

    17

    (0.5 * (17 - 12.5)) + 12.5

    14.75

    Apr

    20

    (0.5 * (20 - 14.75)) + 14.75

    17.375

    May

    22

    (0.5 * (22 - 17.375)) + 17.375

    19.688

    Jun

    20

    (0.5 * (20 - 19.6875)) + 19.6875

    19.844

    Jul

    25

    (0.5 * (25 - 19.844)) + 19.844

    22.422

    Aug

    27

    (0.5 * (27 - 22.422)) + 22.422

    24.711

    Sep

    30

    (0.5 * (30 - 24.711)) + 24.711

    27.355

    Oct

    35

    (0.5 * (35 - 27.355)) + 27.355

    31.178

    Nov

    37

    (0.5 * (37 - 31.178)) + 31.178

    34.089

    Dec

    40

    (0.5 * (40 - 34.089)) + 34.089

    37.044