Trend-Following Strategy — The Classic Multi-Decade Edge

Trend-following is a strategy class that buys assets in price uptrends and sells in downtrends, holding positions through pullbacks until the trend explicitly ends. It is the most academically-validated trading approach — multi-decade studies of CTA (Commodity Trading Advisor) performance show trend-followers earn risk-adjusted returns comparable to equity indices with weaker correlation to broader markets, making them valuable portfolio diversifiers.

The strategy's defining characteristics: low win rate (30-45%), high reward-to-risk ratio (2.5-4:1), low trade frequency (1-5 trades per week per pair), wide stops (typically 80-200 pips), and long holding periods (days to weeks per position). The mathematics favours the strategy on long time horizons but punishes it during ranging regimes where the strategy generates many losing trades waiting for trends that never materialise.

Trend-following demands patience. A trader who enters a trend-following EA expecting consistent monthly returns will be disappointed roughly 30-40% of months. Those losing months are followed by stronger winning months when trends finally develop. Across 12-month windows, healthy trend-following EAs deliver 15-40% net annualised returns. Across 1-month windows, the variance is much higher.

Trend identification typically uses a combination of (1) moving average alignment (e.g. 50-period EMA above 200-period EMA = uptrend on the timeframe), (2) higher-highs-and-higher-lows price structure, (3) ADX or similar trend-strength indicator above a threshold (often 25), and (4) optional multi-timeframe confirmation (H4 trend aligned with D1 trend produces highest-conviction setups).

Entry logic: most trend-followers enter on pullbacks rather than at trend continuation extremes. A typical setup waits for price to retrace to a moving average (often 20-50 EMA on the entry timeframe) or to a Fibonacci retracement level (38.2%-61.8% of the prior swing), then triggers entry on a reversal pattern (bullish/bearish engulfing, pin bar) confirming the pullback has ended.

Exit logic uses one of two approaches: (1) fixed stops at swing high/low with trailing take-profits that move with the trend (typical ATR-based trailing), or (2) trend-reversal exits where the position closes when the trend signal flips (moving average crossover reverses, structural break of trend line). Approach 2 holds positions longer through pullbacks but accepts deeper drawdowns; approach 1 banks profits earlier but sometimes exits before the bulk of the trend's move.

Position sizing: fixed-fractional at 1-2% per trade is standard. Because individual stops are wide (100+ pips), the per-trade risk in dollar terms is substantial — a 1% risk on $10,000 with 100-pip stop on EURUSD = $100 risk, requires 0.1 lot sizing. Trend-followers run multiple uncorrelated pairs to diversify; portfolio-level risk caps at 3-5% across all open positions.

Trend-following has the longest academic record of any retail-accessible strategy. Donchian's turtle-trading experiments in the 1980s, the Andersen-Lim (1989) study of trend persistence in commodities, and the multi-decade CTA index performance data all support the empirical claim that trends in financial markets exhibit positive autocorrelation over multi-day-to-multi-month horizons — long enough for systematic trend-following to extract risk-adjusted alpha.

The classic 'Turtle Traders' system developed by Richard Dennis and William Eckhardt in 1983 codified Donchian Channel breakout trend-following on commodities; the same methodology applied to forex during the 1990s-2010s produced documented profitable track records. Modern academic work (Hurst, Ooi, Pedersen 2014 'A Century of Evidence on Trend-Following Investing') confirms the edge has persisted across asset classes and decades, though it has compressed somewhat as more institutional capital competes.

Current state of forex trend-following: the edge remains positive but margins are tighter than 2000-era backtests suggest. Profit factors of 1.8-2.5 were typical in 2005-2015 backtests; live 2020-2026 operation shows 1.4-1.9 typically. The strategy still produces meaningful returns; the absolute return magnitude has compressed.