Moving averages (SMA and EMA)

If you had to note down every day how much a cup of coffee costs in your favourite café, you would soon see that the price jumps: a promotion one day, an oat milk surcharge the next. To see the real direction, you would calculate the average of the last twenty days. That is exactly what a moving average does: it turns a nervous series of prices into a smooth line from which it is easier to see where the trend is going.

What you will learn

• how the simple (SMA) and exponential (EMA) averages are calculated and how they differ,
• what the lag of an average means and why it is both its main weakness and its strength,
• how to use averages to judge a trend and why they give false signals in a range.

SMA and EMA

The simple moving average (SMA) of n sessions is just the arithmetic mean of the last n closes. Every session has the same weight. A twenty-day SMA "forgets" the price from 21 days ago overnight, which sometimes causes jumps in the average when an unusually high or low price drops out of the window.

The exponential moving average (EMA) gives more weight to newer data. It is calculated recursively:

EMA(today) = α × close + (1 − α) × EMA(yesterday), where α = 2 / (n + 1)

For n = 20 that gives α ≈ 0.095, that is about 9.5% weight for today's price. That is why the EMA reacts faster than an SMA of the same length and has no jumps caused by old data "dropping out".

In practice the most common set is 20, 50 and 200 sessions (or weeks on weekly charts). The first describes short-term mood, the second the medium-term trend, the third the overall direction. The number 200 has no magical properties: it is about ten months of trading, and its power comes mainly from the fact that everybody looks at it.

Lag

Every average lags, because it is based solely on past prices. An SMA of n sessions lags by roughly (n − 1) / 2 sessions: a 50-day average "sees" the market as it looked about 25 days ago. The EMA lags a little less, but it lags too.

That is the main weakness: signals from averages come after the move has already started. But it is also a strength: the lag works as a filter. The average does not react to a one-day jerk, so it ignores the noise that would easily trigger a false alarm. Every average is a compromise between speed and calm. A short one is fast and nervous, a long one calm and late.

How to use averages

• Slope. A rising average suggests an uptrend, a falling one a downtrend, a flat one a consolidation. Sometimes that is all the information you need.
• Position of price. Price above the 200 average is, for many investors, the conventional border between a "healthy" and a "sick" chart.
• Dynamic support. In a strong trend price often pulls back to the 20 or 50 average and bounces off it. The average then acts as a moving support level (lesson 2).
• Crosses. When a short average crosses a long one from below (for example the 50 crossing the 200) we speak of a golden cross, and when from above of a death cross. It sounds ominous, but it is a strongly lagging signal: by the time it appears, the move is usually already under way.

Chart example: Meta Platforms (META), recent years

Meta is a large, liquid stock whose daily chart over the last three years contains both long trends and sharp reversals. Such a chart shows averages at work like in a textbook: in a trend price stays on one side of the averages, and their lag means the reversal signal comes long after the low or the high.

Open the daily chart of META for the last three years and add the 50 and 200 SMAs and the 20 EMA. Find a clear low after which price started to rise, mark when price crossed the 200 average from below after it, and compare that with the date of the actual low. How many percent of the rebound would someone waiting for that signal have missed? And how many bad trades would the person who watched it have avoided? Both answers are true.

How it fails

• Consolidation. When price wanders around the average and the averages are flat and tangled, crosses appear every few days and each is false (whipsaw). It is the most costly scenario for using averages.
• Lag. Cross signals come only after a large part of the move.
• A self-fulfilling number. The 200 average works partly because everybody believes in it. When it stops being fashionable, its strength will fade.
• Parameter fitting. By changing the lengths you will always find a pair of averages that "fitted" history well. That is not proof it will fit the future.

Check yourself

1. What is the smoothing factor of an EMA(n)?
2. Which average reacts faster, SMA or EMA, and why?
3. By how many sessions does an SMA(n) lag?
4. Why do averages give false signals in a consolidation?
5. What does a golden cross mean?

Next lesson

Lesson 8. Timeframes and top-down analysis: the same chart can shout "buy" and "run" at the same time, depending on the window you look through. You will learn three windows that turn chaos into a calm, readable picture.

Educational material, not investment advice.