The moving average is the oldest filter in technical analysis: the average of the last n closes, recomputed on every bar. Raw price wobbles even when nothing meaningful is happening; the average cancels most of that back-and-forth and leaves the underlying direction visible. Almost everything you will meet in this catalogue — MACD, Bollinger Bands, Supertrend, ADX — is built on top of this idea.
In plain terms — Price is a crowd shouting different numbers every second; the moving average is the crowd's average voice. It does not tell you the next number: it tells you which way the chorus is drifting.
What it measures
A moving average answers a single question: noise aside, where is price going? The parameter n (the window) decides how much noise is removed — and how much lag you pay in exchange:
| Window | Behaviour | Typical use |
|---|---|---|
| Short (9–21) | Responsive, jumpy | Riding the current trend, dynamic support |
| Medium (50) | The compromise | Swing-trading context |
| Long (100–200) | Stable, slow | Regime filter: above or below? |
The trade-off is structural, not a flaw to engineer away: more smoothing means more lag. An n-period SMA lags by roughly n/2 bars; no parameter choice removes that cost — it only moves it around.
How to read the chart — Thin line: price. Gold line: the EMA 20, riding the trend and touched by pullbacks. Teal line: the SMA 50, slower, defining the context. Interactive — hover (or tap) the highlighted points: the pullback to the EMA, the role of the SMA 50, structural lag.
How it is calculated
SMA (Simple Moving Average) — the arithmetic mean of the last n closes:
SMA(5) on closes 100, 102, 101, 104, 103 → (100+102+101+104+103) / 5 = 102
On the next bar the window slides: the new close enters, the oldest drops out.
EMA (Exponential Moving Average) — weights recent bars more heavily, with exponentially decaying weights:
EMA(today) = α · Close(today) + (1 − α) · EMA(yesterday), with α = 2/(n+1)
With n = 20, α ≈ 0.095: each new bar carries about 9.5% of the updated value. The practical result: at equal periods the EMA turns earlier at reversals and trembles more in noise. Neither is "better": they are two settings of the same trade-off. Common conventions: EMA on short windows (9, 12, 21), SMA on long ones (50, 100, 200).
Reading it in practice
- Regime filter — price above the SMA 200: a context for hunting longs; below it: a defensive context. It is not an entry signal — it answers "what market am I in?", the question that comes before every other.
- Dynamic support and resistance — in orderly trends price stretches away from the average and returns to it; the touch of the EMA 20/50 is among the most watched spots in trading. Treat it as an attention zone, to be confirmed by the price reaction.
- Building block for other tools — the MACD is the distance between two EMAs; Bollinger Bands are an SMA with standard-deviation margins; the Supertrend is an ATR band with average-like logic. Understand the moving average well and the rest of the catalogue reads itself.
In the cyclic analysis tradition the average is also used centred: shifted back by half its window it stops lagging and becomes a dissection tool for the waves that make up price.
Limits and traps
Warning — crossovers in sideways markets. "Buy when the fast average crosses the slow one" only works where an underlying direction exists. In a sideways market the averages keep intertwining and crossovers produce serial false signals (whipsaw): you buy high and sell low, several times in a row. Before trading a crossover, verify that a trend exists — that is the job of the ADX.
- Lag cannot be removed, only chosen. "Magic" parameters found in hindsight are almost always overfitting.
- The average is computed on closes: intrabar events (spikes, wicks) pass straight through it without leaving a trace.
Links
- moving-average — the encyclopedia concept
- macd — two EMAs subtracted
- bollinger-bands — an SMA 20 with deviation bands
- adx — measures whether the trend exists
- indicatori — catalogue hub