Trend Following
How does a moving average crossover strategy work?
A moving average crossover trades in the direction a faster moving average crosses a slower one, treating the cross as a signal that the trend regime has shifted. The rules below cover choosing the pair of averages, entry and exit, position management, and why the signal always arrives after the actual turn.
How it works
It trades the cross, not the trend itself
A moving average smooths price over a fixed lookback. Plot two — one faster, one slower — and the point where the faster line crosses the slower one marks a shift in the average trend: price has, on average, moved further from where it sat over the longer window than the shorter one. The strategy buys when the fast average crosses above the slow one and sells or shorts when it crosses below, treating each cross as a new trend regime rather than trying to call the top or bottom directly.
- 01
Choose a fast and slow average pair and fix it
Common pairs are a 10/50, 20/50, or 50/200 combination of simple or exponential moving averages. The specific lengths matter less than picking a pair and keeping it fixed — changing the lengths after a bad signal is curve-fitting to the past, not a rule.
- 02
Enter on the cross, not before it
Go long when the fast average crosses above the slow average, and go short or exit a long when it crosses below. Anticipating the cross by entering on the approach gives up the confirmation the rule is built to require.
- 03
Place the stop below recent structure, not at a fixed distance
A stop placed under the most recent swing low (for a long) or above the most recent swing high (for a short) ties the risk to what the market has actually done, rather than to an arbitrary point count.
- 04
Exit on the opposite cross or a structural stop, whichever comes first
The position is held until either the averages cross back the other way or the structural stop is hit. There's no fixed holding period — a crossover system can hold a position for a long stretch if the trend keeps running.
Position sizing
Risk is set by the stop distance, which moves with volatility
Because the stop sits at a recent swing point rather than a fixed distance, the position size has to adjust to keep dollar risk constant. A volatile stretch produces wider swings and a wider stop, which calls for a smaller position at the same percentage risk; a quiet stretch does the opposite. Sizing off a fixed share count instead of the stop distance means the actual risk taken varies from trade to trade without the trader choosing that.
Common mistakes
Where this setup usually goes wrong
- 01
Using too short a lookback and trading every wiggle
A fast pair like a 5/10 crosses often in a choppy market, generating a signal on noise rather than an actual regime change. The shorter the averages, the more the strategy needs a genuinely trending market to avoid getting whipsawed.
- 02
Overriding the exit because the trend 'feels' like it will continue
Holding past the opposite cross on a hunch reintroduces the discretion the rule was built to remove, and it's usually done after the trend has already stalled, not while it's strong.
- 03
Ignoring how late the signal actually arrives
By construction, a crossover only fires once the fast average has moved enough to cross the slow one — after a real part of the move has already happened. Expecting an early entry from a lagging indicator sets up disappointment with a rule that was never designed to give one.
- 04
Trading the same pair across every instrument without checking it
A pair tuned to a slow-moving index doesn't necessarily fit a fast-moving single stock or a thin futures contract. The lengths that produce a reasonable signal frequency differ by how much the instrument typically moves.
Limitations
It gives back a chunk of every move and does poorly in a range
A crossover always lags the actual turn, since the fast average has to move enough to cross the slow one first — some of the early move is given up on entry, and some of the profit is given back before the exit signal confirms the trend has ended. In a sideways, range-bound market, the two averages cross back and forth repeatedly with no sustained trend to capture, producing a series of small losses. This strategy needs a market that actually trends for a meaningful stretch to be worth running.
Backtest a crossover pair before trading it live
The right fast/slow pair depends on the instrument and timeframe you actually trade. Backtesting on Traders Journal runs a moving average crossover against historical price data so you can see how a specific pair would have performed before committing to it.
Explore backtestingQuestions
Which moving average crossover is most common: 50/200 or 20/50?
The 50/200 pair — often called the golden cross going up and the death cross going down — is the most widely referenced on daily charts and trades infrequently, aiming for major trend shifts. A 20/50 pair reacts faster and trades more often, better suited to shorter swing timeframes. Neither is objectively better; they trade off signal frequency against lag.
Should I use simple or exponential moving averages for a crossover?
An exponential moving average weights recent price more heavily, so it turns and crosses somewhat sooner than a simple moving average of the same length. That makes exponential pairs a bit more responsive and simple pairs a bit steadier — a real trade-off, not a settled answer, and worth testing against the specific instrument you trade.
Why did the moving average crossover give a signal after the move already happened?
That's a property of the rule, not a malfunction. A crossover only fires once the faster average has moved enough to cross the slower one, which by definition happens after the price shift that caused it. The strategy trades confirmed regime changes, not early turns.
Can a moving average crossover be combined with other indicators?
Yes — a common addition is an RSI or volume filter to skip crosses that happen with weak momentum behind them, or a longer-term trend filter to only take crosses in the direction of the bigger trend. Any addition changes the rule and should be tested as its own version rather than assumed to be an improvement.