BUILD: COURSE 1 | LESSON 2
Moving averages and crossovers
Learning objectives
Explain how SMAs and EMAs are calculated and why every moving average lags by design
Choose settings (20/50/200, 9/21) understanding the trade-off between responsiveness and noise
Use crossovers and the 200-day MA as context filters rather than standalone entry signals
What a moving average actually is
A moving average (MA) answers one question: what has the average price been over the last N periods? Plot that answer on every candle and you get a smooth line that filters the noise out of price.
The simple moving average (SMA) is exactly what it sounds like. A 20-day SMA adds the last 20 closing prices and divides by 20. If EURUSD closed between 1.0700 and 1.0800 over the past month, the 20-day SMA sits somewhere in the middle, around 1.0750. Each new day, the oldest close drops out and the newest drops in.
The exponential moving average (EMA) weights recent prices more heavily. The formula:
For a 20-period EMA, k = 2/21 ≈ 0.095 — today's close gets about a 9.5% say in the new value, and everything before gets the remaining 90.5%, with influence decaying exponentially into the past. Intuitively: the SMA is a committee where every one of the last 20 days gets an equal vote; the EMA is a committee where yesterday shouts loudest and the distant past whispers.
The consequence: EMAs turn faster, SMAs are steadier. Neither is "better" — they sit at different points on the same trade-off, which brings us to the one property you must internalise: every moving average is a summary of the past. An MA cannot lead price; it can only lag it. When price crashes, the MA follows it down late. That lag is not a flaw to engineer away — shorten the period and you get less lag but more false wiggles. Lag versus noise is the whole game.
Standard settings and what they're for
You'll meet the same handful of settings everywhere, and there's value in that: levels get self-fulfilling weight because thousands of traders and algos watch the same lines.
- 200-day SMA — the institutional dividing line between bull and bear regimes. "Above the 200-day" is shorthand for a healthy long-term trend on everything from S&P 500 CFDs to gold.
- 50-day SMA/EMA — the medium-term trend; the 50/200 relationship defines the famous crossovers below.
- 20-period — the short-term trend and the default "dynamic support" line traders watch on pullbacks (also the midline of Bollinger Bands, as you'll see in Lesson 4).
- 9 and 21 EMA — the fast pair popular with intraday and swing traders; the 9/21 crossover is a common momentum trigger.
Trade-offs, stated honestly. Shorter settings (9/21) get you into moves earlier and whipsaw you constantly in ranges. Longer settings (50/200) rarely whipsaw but surrender a large slice of every move at both ends — by the time the 200-day confirms a trend change, the move may be 15–20% done. There is no setting that is both early and reliable. Anyone selling you one is selling you curve-fitting: settings optimised to perfection on last year's data routinely fall apart on next year's.
Crossovers: golden crosses, death crosses, and their real hit rate
A crossover fires when a faster MA crosses a slower one. The celebrity versions:
- Golden cross: 50-day crosses above the 200-day — bullish regime signal.
- Death cross: 50-day crosses below the 200-day — bearish regime signal.
Worked example. Suppose gold (XAUUSD) has ground down from 2,430 to 2,310 over three months; the 50-day (say 2,352) sits below the 200-day (2,368). Price then bases and rallies to 2,395. The 50-day curls up and crosses the 200-day at around 2,370. Notice what the signal did: it confirmed a recovery that was already 85 dollars old. That's the lag tax, and you always pay it.
Now the honest numbers. Long-run studies of golden/death crosses on equity indices show they capture big trends and sidestep major bear markets — but the raw signal hit rate is unimpressive, commonly cited near 50–60%, and in choppy, sideways markets crossovers lose repeatedly as the averages braid around each other. A 9/21 EMA cross on a ranging H1 chart can fire a dozen times a week, nearly all of them noise. Crossover systems make their money from a few large winners paying for many small losers — which only works if you actually take the small loss every time. The signal doesn't carry the edge; the asymmetry between winners and losers does. That's risk management again, wearing an indicator costume.
Two upgrades that materially improve crossover use:
- Trade them only with the higher-timeframe trend. Take 9/21 bullish crosses only while price is above the daily 200-SMA. You'll skip most of the chop.
- Use the slow MA as your regime filter, not your trigger. Many professionals never trade the cross itself — they simply refuse longs below the 200-day and refuse shorts above it, and get their entries from structure (Lesson 1) or momentum (Lesson 3).
Dynamic support, MA slope, and the mistakes to avoid
Beyond crossovers, MAs earn their place in three quieter ways:
- Slope = trend health. A rising 50-day under rising price is a trend behaving; a flat 200-day tells you you're in a range and trend tools will underperform. When all your MAs are flat and braided, the correct signal is often no trade.
- Dynamic support/resistance. In steady trends, pullbacks repeatedly find buyers near the 20 or 50 EMA — e.g. USDJPY trending from 152 to 157 might tag the daily 21-EMA four times, each tag a lower-risk entry zone than chasing the highs. Treat the MA as a zone, not a line price must respect to the pip — it won't.
- Distance from the MA = stretch. When price is 3% above its 20-day MA on a pair that normally strays 1%, you're late. Mean reversion doesn't owe you anything, but chasing statistical stretch is how losing streaks start.
And the classic mistakes: stacking five MAs on one chart (they'll contradict each other and you'll cherry-pick); re-optimising settings after every losing week (curve-fitting); and treating a single MA touch as a signal with no stop and no structure behind it. One tool, one job: moving averages define the regime. Something else must define the trade.
Key takeaways
An SMA weights all N periods equally; an EMA (k = 2/(N+1)) weights recent prices more — faster to turn, more prone to noise. Both lag by construction.
Standard settings (20, 50, 200; 9/21 EMA) matter partly because everyone watches them; there is no setting that is both early and reliable.
Golden/death crosses confirm trends late, with modest raw hit rates (~50–60%); crossover systems profit from asymmetric winners, not accuracy.
The strongest everyday use of MAs is as filters: trade with the side of the 200-day, read slope for regime, use the 20/50 EMA as pullback zones.
Flat, braided MAs are a signal too — the signal to stand aside. No moving average setup works without a stop and sizing behind it.