Moving averages explained: SMA vs EMA, and does the crossover work?

A moving average is the average price over the last few candles, recalculated on every new candle so that it moves along the chart. It smooths out the noise so the direction of the recent trend is easier to see. The two common kinds are the simple moving average (SMA), where every candle counts equally, and the exponential moving average (EMA), which gives the latest candles more weight and therefore reacts faster.

The most popular way to turn moving averages into a signal is the crossover: when a fast average crosses above a slow one it is read as a buy signal, and below as a sell signal. We measured how that signal actually performed. Across our 41 markets, hourly EMA 9/21 crossovers worked 49.2% of the time - about the same as a coin toss. The chart, scoreboards and the streak comparison below show what that looks like in practice.

See it on a real chart: Gold (XAUUSD) liquidation heatmap — the modeled clusters above and below the current price.

How a simple moving average (SMA) is calculated

Add up the closing prices of the last N candles and divide by N. A 20-period SMA on the 1-hour chart is the average close of the last 20 hours. When a new candle closes, the oldest one drops out and the newest one comes in, so the line moves.

Because every candle has the same weight, an SMA is smooth but slow. A sharp move only shows up gradually, and an old price spike has as much influence as yesterday's close until it drops out of the window - which can make the line jump for no current reason.

How an exponential moving average (EMA) is calculated

An EMA keeps a running value and moves it a fixed fraction towards each new close. The fraction is 2 / (N + 1): for a 9-period EMA that is 20% of the gap between the last EMA value and the new close, for a 21-period EMA about 9%. Recent candles therefore matter most, and older ones fade out gradually instead of dropping off a cliff.

The result is a line that hugs price more closely and turns sooner than an SMA of the same length. The price of that speed is more false turns in a sideways market.

SMA or EMA: which one?

• The EMA reacts faster; the SMA is smoother. Neither is more correct - they answer slightly different questions about the same prices.
• Short periods (9, 10, 20) follow price closely and turn often. Long periods (50, 100, 200) move slowly and describe the broader direction.
• Well-known pairs: 9/21 EMA on intraday charts, 50/200 SMA on daily charts (the 50 crossing above the 200 is the widely quoted golden cross, below it the death cross).
• The lengths are conventions, not discoveries. There is no period that is right for every market or timeframe.

How traders use moving averages

• As a trend filter: price above a rising average is read as an uptrend, below a falling one as a downtrend.
• As a crossover signal: the fast average crossing the slow one, which is what we tested below.
• As dynamic support or resistance: some traders watch how price behaves when it comes back to an average.
• As a reference for distance: how far price has stretched away from its average.

All of these describe what price has already done. A moving average is built entirely from past closes, so by construction it lags: a crossover happens after the move that caused it.

What we measured: EMA 9/21 crossovers on real charts

We take every crossover of the 9-period EMA over the 21-period EMA (buy) and under it (sell), on the chart of each market, and check what price did next: a signal counts as a win if price moved one average true range (ATR) in its direction before it moved one ATR against it, within 48 candles, and as a loss if the reverse happened first. Nothing is optimised and nothing is hidden; the same rule is used for every market and every indicator in this series.

The chart below is gold on the 1-hour chart, with both averages drawn exactly as we compute them and every crossover in the window marked.

Why a winning streak does not mean the signal is working

Look at the two rows of W and L above. Both contain runs - four, five, sometimes more wins or losses in a row. One of them is a coin. Streaks are exactly what a random sequence produces: in 100 tosses of a fair coin, a run of six or seven of the same side is normal. A trader who meets the indicator during a good run concludes it works; one who meets it during a bad run concludes it is broken. Both are looking at chance.

We tested the obvious follow-up question, the roulette one: after several losses in a row, is the next signal more likely to win? Across 15,192 hourly signals on 41 markets, after one to five losses in a row the next signal worked 41-56% of the time - within the range chance alone produces. A run tells you what already happened, not what comes next.

What that means in practice

• A moving average is a useful description of recent direction. On our data, the crossover on its own is not a forecast.
• Judge any signal by many results against a fair baseline, never by the last few. Twenty signals are not enough to tell 55% from 45%.
• The same logic applies to anyone showing you a streak of winning trades from an indicator. Ask how many signals, against what baseline.

Every market page shows this scoreboard for its own chart, on four timeframes. For gold, see the XAUUSD signal truth section. The method, and what we found when we tested our own levels the same way, is on the methodology page.