Trend Following in Commodities: Turtle Traders and Donchian Breakouts

Trend following buys markets making new highs and sells markets making new lows, then holds while the move continues; it is the oldest systematic approach in commodities. The best-known version is the Turtle traders' Donchian channel breakout from the 1980s: enter on a 20- or 55-day breakout, size each position by volatility, and exit on a shorter breakout the other way. It typically loses on more trades than it wins and depends on a few large trends, so it needs discipline, diversification and patience.

Why commodities trend

• Slow supply response. When prices rise, it takes years to open mines and fields and at least a season to plant more crops, so shortages persist and prices keep climbing. When prices fall, producers often keep pumping or mining to cover costs, so gluts persist too.
• Inventory feedback. Low inventories push the curve into backwardation, which gives long positions positive roll yield, and high prices eventually ration demand. These processes take months.
• Gradual information and herding. News about harvests, demand and policy arrives piece by piece, and investors react at different speeds.
• Hedging pressure. Producers and consumers hedge in predictable ways that can add to moves.

Academic research supports the idea. Moskowitz, Ooi and Pedersen's 2012 study "Time Series Momentum" found that a market's own past returns tended to predict its future returns across a broad set of futures markets, including commodities. A later paper by Hurst, Ooi and Pedersen examined trend-following over roughly a century of data. Both also show that results vary by period, with long stretches of flat or poor performance.

Richard Donchian and the channel breakout

Richard Donchian, often called the father of trend following, ran one of the first public managed futures funds and popularised simple breakout rules, such as buying when price exceeds its highest level of the previous four weeks. The Donchian channel, the highest high and lowest low of the last N days, is named after him.

The Turtle experiment

In 1983 the commodity trader Richard Dennis and his partner William Eckhardt disagreed about whether trading could be taught. Dennis recruited a group of trainees, the "Turtles", taught them a complete rule-based system, and gave them money to trade. The rules were later published by former Turtles and others. In summary:

• N: the 20-day average true range, a measure of volatility.
• Unit size: 1% of equity divided by (N x dollars per point), so a 1-N move changes equity by about 1%.
• System 1: enter on a 20-day breakout (skipping the signal if the previous System 1 breakout would have been a winner); exit on a 10-day breakout the other way.
• System 2: enter on a 55-day breakout; exit on a 20-day breakout the other way.
• Stop: 2N from entry.
• Pyramiding: add one unit for every 1/2 N move in your favour, up to 4 units per market, moving all stops to 2N from the latest entry.
• Portfolio limits: maximum 6 units in closely correlated markets, 10 in loosely correlated markets and 12 in one direction.

Long trades vs short trades

In commodities, the long and short sides of a trend system tend to behave differently. In general terms:

• Upside moves can be sudden. Supply shocks, weather events and wars can cause sharp spikes, which favour longs already in position and hurt shorts badly.
• Downtrends often grind. Gluts build gradually as inventories rise, so falls can be slower and choppier.
• Roll yield differs by side. Tight markets in backwardation reward longs; oversupplied markets in contango reward shorts.
• Short squeezes. Crowded short positions can be forced to cover all at once, as the nickel squeeze in 2022 showed.

These are tendencies, and the balance varies by market and period. The practical rule is to test the long side and the short side separately rather than assume they are equal. The DanioX commodity backtest page lists long and short trades for several well-known strategies, including the Turtle rules and a 20-day trend breakout, so you can see how the two sides behave instead of taking it on trust. DanioX also runs a related 20-day breakout for crypto, explained on the trend signals page.

Why trend following is hard to live with

• Many small losses. In sideways markets, breakouts fail repeatedly ("whipsaws").
• Giving back profits. Exits wait for a 10- or 20-day reversal, so a big chunk of open profit is returned at the end of every trend.
• Long drawdowns. Months or even years can pass with little progress, and the next big trend may come in a market you were tempted to drop.
• It needs many markets. You cannot know which market will trend, so you need a spread of them, sized by volatility, as the diversification lesson explains.

How to test a breakout system yourself

1. Choose a universe of liquid markets across metals, energy, grains, softs and livestock.
2. Use adjusted continuous daily data and decide how you handle rolls.
3. Write exact rules: entry, initial stop, exit, sizing and portfolio limits.
4. Include realistic costs: commissions, slippage and roll costs.
5. Fix the rules before testing, then evaluate on years and markets you did not use to design them.
6. Look at results by market, by side and by year, not just the total.

The backtesting lesson covers each step in detail, and the methodology page explains how DanioX approaches its own tests.