How to Backtest Forex Strategies on Historical Data

Forex is the largest financial market on the planet — $7.5 trillion in daily volume. Unlike crypto, it has decades of clean historical data, well-understood microstructure, and predictable session patterns. This makes it arguably the best market for systematic backtesting.

I've been backtesting forex strategies for over a decade, going back to my early days building automated systems for MetaTrader. The tools have evolved dramatically, but the principles haven't changed.

Forex Data: Quality Matters More Than Quantity

The quality of your forex data directly determines the quality of your backtest. There are several sources with different trade-offs:

Tick data from providers like Dukascopy offers the highest resolution — every price change with exact timestamps. For scalping strategies and strategies with tight stops, tick data is essential. The downside: file sizes are enormous. EUR/USD tick data for one year can exceed 10 GB.

1-minute OHLCV data is the practical sweet spot for most strategies. It captures intrabar movement that daily data misses, without the storage overhead of tick data. Ten years of 1-minute data for 27 major pairs fits in under a gigabyte.

Daily data is free and widely available but insufficient for any strategy with intraday timing. If your strategy includes "buy at London open" or "sell at New York close," daily bars won't capture this correctly.

Forex-Specific Backtest Configuration

One critical detail many traders miss: spreads aren't constant. EUR/USD spreads widen from 0.5 pips during London session to 2-3 pips during the Asian session's quiet hours. During news events (NFP, ECB decisions), spreads can spike to 10-20 pips. A backtest using fixed 1-pip spreads dramatically underestimates costs during volatile periods.

Session-Aware Testing

Forex has three major trading sessions: Asian (Tokyo), European (London), and American (New York). Each has distinct characteristics:

• London session (07:00-16:00 UTC): Highest liquidity, tightest spreads, strongest trends. ~35% of daily volume.
• New York session (12:00-21:00 UTC): High volatility, especially during London/NY overlap (12:00-16:00). ~25% of daily volume.
• Asian session (23:00-08:00 UTC): Lower volatility, wider spreads, range-bound price action on most pairs. JPY pairs are the exception.

If your strategy works during all sessions equally, that's a yellow flag. Most genuine forex edges are session-specific. A breakout strategy that works during London open but fails during Asian session makes structural sense — liquidity returns and prices discover new levels.

Common Forex Backtesting Mistakes

Ignoring swap rates. If you hold EUR/USD short overnight, you earn swap interest. Hold it long, you pay. Over months, swap can add or subtract 3-5% from your returns. Any strategy holding positions for days or weeks must model swaps.

Testing on Sunday/Monday gaps. Forex closes Friday 21:00 UTC and reopens Sunday 21:00 UTC. The gap can be 20-100 pips on major pairs after weekend events. Strategies with open positions over weekends must account for this gap risk.

Using the wrong pip value. JPY pairs have 2-decimal pricing (1 pip = 0.01). Most other pairs use 4-decimal (1 pip = 0.0001). A strategy that works "1 pip stop-loss" on EUR/USD is risking 10x less than the same setup on USD/JPY if pip values aren't normalized.

"The forex market rewards patience and punishes complexity. My best strategies use two indicators and simple rules. My worst strategies used seven indicators and felt brilliant." — From my trading notes, 2019

Multi-Pair Validation

A robust forex strategy should work across correlated pairs. If your EUR/USD strategy is profitable, test it on GBP/USD and AUD/USD (positively correlated) and on USD/CHF (negatively correlated). If it works on all four, you've likely found a genuine pattern in USD behavior rather than an artifact of EUR/USD data.

The 27 major and minor forex pairs provide an excellent testing universe. Starting with the backtesting fundamentals and applying them to currency markets gives you a massive advantage over traders who test on a single pair.