MACD Crossover strategy: rules and backtest
MACD Crossover trades crosses of the MACD line and its signal line, catching the change in the rate at which the moving averages diverge.
| Parameter | Value |
|---|---|
| Type | Counter-trend |
| Timeframe | M1–M5 |
| Complexity | Pro |
| Instrument | GBP/USD |
How the signal works
A cross of MACD and its signal line is the point where the speed at which the fast and slow EMAs diverge changes sign. The strategy enters in the direction of the new cross and holds until the opposite one, staying in the market almost continuously.
Because it is constantly in the market, the strategy is sensitive to ranges: there the lines cross often, and each cross turns out to be false. MACD is not normalised, so tuning thresholds is pointless — only the fact of the cross works.
Verification on real data
This strategy’s rule is run on real quotes with no parameter fitting to history. The rule tested was “MACD Crossover: MACD crossing its signal line”:
- Long when the MACD line crosses above its signal line.
- Short when it crosses below.
- The position is held until the opposite crossing.
Pros and cons
- Reacts to a change in momentum before price turns.
- Requires no threshold tuning — the signal is binary.
- Works well in pronounced trends.
- Almost always in the market — it accumulates costs on frequent reversals.
- In a range it produces a stream of false crosses.
- Double smoothing means it lags on sharp reversals.
Pitfalls
MACD Crossover without a trend filter is a classic commission-generating machine in a ranging market: the lines cross on every swing, and the strategy whipsaws back and forth. The second error is comparing the magnitude of MACD across periods: the indicator is not normalised, and only the side of the cross carries meaning.
Who it suits
For traders of trending instruments who accept a high trade frequency and its costs. On an instrument that spends most of its time in a range, the strategy will be unprofitable.
Frequently asked questions
How does the signal-line cross differ from a zero-line cross?
The signal-line cross happens earlier and marks a change in the rate at which the averages diverge. The zero-line cross is a later, rarer event: it means the fast EMA has drawn level with the slow one.
Why is the strategy almost always in the market?
The rule holds a position from one cross to the opposite one; there is no intermediate "out of market" state. Hence the high exposure and sensitivity to costs.
Can the number of false signals be reduced?
Yes, by adding a market-state filter — for example, entering only when ADX is above a threshold. The base version has no filter, so its backtest on a ranging sample shows a weak result.