TradeAlmanac
Sign in

Exponential moving average

An average in which the weight of an observation decays as it recedes into the past, so recent intervals count for more.

Formula

EMA_t = \alpha P_t + (1 - \alpha) EMA_{t-1}

Alpha is the smoothing coefficient derived from the chosen period; P is the price of the interval.

How to read the number

It answers to fresh data faster than a simple average because old observations are not dropped all at once when they leave a window — they lose weight gradually.

When the metric lies

The quicker reaction is paid for in false turns of the line. The past never leaves the calculation entirely, so the lag never disappears either.

Also known as: ema

Related terms