Exponential Moving Average

02 / Explanation

A small numerical example

For a three-period EMA, the conventional smoothing factor is 2 / (3 + 1) = 0.5. If the previous EMA is €102 and the current closing price is €108:

EMA = 0.5 × 108 + (1 − 0.5) × 102 = 105

On the next period, if the price is €111, the new calculation uses €105 as its starting EMA: 0.5 × 111 + 0.5 × 105 = 108.

How the indicator is calculated

1. Select the period parameter n and calculate the smoothing factor α.
2. Establish an initial value, commonly an SMA of the first n observations.
3. Multiply the newest observation by α.
4. Multiply the preceding EMA by (1 − α) and add the two parts.
5. Carry the result forward as the preceding EMA for the next period.

Because the calculation is recursive, all earlier data can influence the current value, but that influence becomes progressively smaller.

Practical application

EMA is often chosen when an analyst wants smoothing that responds relatively quickly to new observations. It can support trend monitoring, short- and long-period crossover analysis, or comparisons between current values and a recently weighted reference level. The method also applies beyond prices wherever gradual decay is more suitable than an abrupt rolling-window cutoff.

Advantages

• Recent observations have more influence than distant ones.
• Its response is smoother than the raw series but generally quicker than an equivalent SMA.
• The recursive update requires only the newest observation and previous EMA.
• Older observations fade gradually instead of leaving the calculation suddenly.

Limitations

• The initial value affects the early part of the series.
• Greater responsiveness can also make short-period EMA more sensitive to noise.
• The conventional period label is an interpretation of the smoothing factor, not a strict window.
• Like other moving averages, EMA is backward-looking and is not a forecast on its own.

Comparison with SMA and WMA

Compared with the simple moving average, EMA places more emphasis on recent observations and has no abrupt window boundary. Compared with the weighted moving average, EMA uses an exponential decay rule and carries information recursively, whereas WMA normally applies an explicit finite set of weights. EMA is convenient for continuous updating; WMA offers more direct control over the weight pattern.