Home/Blog/Backtesting Forex Moving Average Crossover Systems: Are They Dead?
Forex9 min read

Backtesting Forex Moving Average Crossover Systems: Are They Dead?

By BacktestEverything·January 6, 2026

# Backtesting Forex Moving Average Crossover Systems: Are They Dead?

The moving average crossover is the most basic trend-following signal in existence: buy when the fast MA crosses above the slow MA, sell when it crosses below. Once highly profitable in forex markets, many traders believe these systems have been arbitraged away. We tested exhaustively to find the truth.

The Exhaustive Test

We tested every combination of fast MA (5 to 50, step 5) and slow MA (20 to 200, step 10) on eight major forex pairs from 2010 to 2025. Thats 180 parameter combinations across 8 pairs, totaling 1,440 backtests. We used both simple and exponential moving averages.

The Sobering Results

Of 1,440 combinations, only 23% produced positive returns after spread costs (1.5 pips average). Only 8% produced a Sharpe ratio above 0.3. And only 2% produced a Sharpe above 0.5. The vast majority of moving average crossover systems are no longer profitable in forex.

The Survivors

The combinations that survived shared characteristics: relatively slow parameters (fast MA 20-40, slow MA 100-200), applied to trending pairs (AUD/JPY, GBP/JPY, EUR/USD), and using exponential rather than simple averages. The best single combination was 30/150 EMA on GBP/JPY with a Sharpe of 0.58.

Why Most Failed

Two primary causes of failure emerged. First, the proliferation of algorithmic trading has compressed the post-crossover moves that once lasted days or weeks into hours. Second, the post-2015 forex market has seen lower trends and higher chop, with many pairs spending extended periods in ranges that generate repeated false signals.

Transaction Cost Sensitivity

Faster MA combinations (like the popular 5/20) generated 40-60 trades per year per pair. At 1.5 pips per round trip, transaction costs consumed 2-4% annually. Slower combinations (like 30/150) generated only 6-10 trades per year, making transaction costs negligible but also providing fewer opportunities.

Adding Filters Helped Marginally

We tested adding an ADX filter (only trade when ADX above 20, indicating trend strength). This improved win rates from 34% to 41% on the 30/150 system and eliminated many whipsaw trades during ranging markets. The overall improvement was approximately 0.1 Sharpe ratio units, meaningful but not transformative.

The Portfolio Approach

Rather than trading one pair with one MA combination, we built a portfolio of 4 pairs (GBP/JPY, AUD/JPY, EUR/USD, USD/CAD) each using their individually optimized MA parameters. The diversified portfolio produced a Sharpe of 0.52, more consistent than any single pair, and a smoother equity curve with 12% maximum drawdown.

Comparison to Modern Alternatives

We compared the best MA crossover system against a simple momentum strategy (buy pairs with positive 3-month return, sell those with negative). The momentum strategy produced a Sharpe of 0.61 versus 0.52 for the MA crossover. More sophisticated trend filters have largely superseded the humble crossover in professional systematic forex trading.

Are They Dead? Not Quite

Moving average crossovers are not dead, but they are on life support. They work only in specific pair and parameter combinations, require filters to be viable, and produce modest returns that trail more sophisticated approaches. For educational purposes and as a foundation for more complex systems, they retain value. As a standalone profitable strategy in modern forex markets, they are no longer sufficient.

Recommendations

If you insist on trading MA crossovers in forex: use slow parameters (30+ fast, 100+ slow), focus on JPY crosses and major pairs with trending characteristics, add an ADX or volatility filter, trade a diversified portfolio of pairs, and have realistic expectations of 3-6% annual returns after costs. For most traders, the effort-to-reward ratio suggests exploring more modern approaches to forex trend-following.

Want to See More Backtests?

Watch our video breakdowns with real data and analysis

Watch Videos

More Articles