Turtle Trading System: Full Rules

At a glance

Origin: A 1983 experiment testing whether trading can be taught
Core signal: 20-day and 55-day Donchian channel breakouts
Position sizing: Volatility units based on N, an early ATR
Lasting contribution: Risk normalisation across markets, not the entry rule

Key takeaways

The Turtles traded a straightforward breakout system; what made it work was the position sizing, the pyramiding rules, and the portfolio-level risk limits.
N, the 20-day average true range, normalised risk so that every market contributed roughly the same volatility to the portfolio.
The system deliberately took every signal, including the ones that looked bad, because skipping signals removes the rare large winners.
The specific parameters are less important than the structure; the breakout lengths would be curve fitting if adopted uncritically today.
The experiment’s real result was behavioural: identical rules produced very different outcomes depending on whether traders followed them.

What the experiment was

In 1983, commodity trader Richard Dennis wagered with his partner William Eckhardt that trading could be taught. He recruited a group of novices, trained them for two weeks in a specific rule set, funded them, and let them trade. Several went on to produce strong records over the following years.

The rules later became public. Their historical importance is not that they are the best trend-following rules, but that they are a complete, self-consistent system covering entries, exits, sizing, pyramiding, and portfolio limits, at a level of specification that most published strategies still fail to reach.

The rules in full

Markets: Liquid futures across currencies, rates, metals, energy, and agriculture. Diversification across uncorrelated markets was structural, not optional.
N (volatility unit): A 20-day exponential average of the true range, functionally the modern ATR. Everything else is expressed in units of N.
Unit size: Unit = (1% of account equity) / (N x dollars per point). One unit therefore represents a 1 percent move in equity for a 1 N price move.
System 1 entry: Buy a 20-day high breakout, but skip the signal if the previous 20-day breakout would have been a winner.
System 2 entry: Buy a 55-day high breakout, always taken, with no skip rule. This ensured the largest trends were never missed.
Adding units (pyramiding): Add one unit for every further 0.5 N of favourable movement, up to a maximum of 4 units in one market.
Stop: 2 N below the entry price of each unit. When units are added, earlier stops are raised so the whole position risks no more than 2 N from the most recent entry.
System 1 exit: A 10-day low for longs.
System 2 exit: A 20-day low for longs.
Portfolio limits: Maximum 4 units per market, 6 units in closely correlated markets, 10 units in loosely correlated markets, and 12 units total on one side of the portfolio.
Drawdown rule: Reduce unit size by 20 percent for every 10 percent of account drawdown, and restore it only after recovery.

The position sizing, which is the real lesson

N              = 20-day average true range, in price points
Dollar volatility = N x dollars per point of the contract
Unit           = (0.01 x Account equity) / Dollar volatility

Example: 1,000,000 account
  Market A: N = 0.0075, point value 100,000  -> dollar vol 750
            Unit = 10,000 / 750 = 13 contracts
  Market B: N = 1.20,   point value 1,000     -> dollar vol 1,200
            Unit = 10,000 / 1,200 = 8 contracts

Both positions now risk approximately the same amount for a
1 N move, even though the markets have completely different
prices, tick sizes, and volatilities.

Turtle unit sizing, still the standard approach in modern systematic trading.

This normalisation is the reason a single portfolio could hold currencies, bonds, and agricultural commodities simultaneously without one market dominating the risk. Every modern systematic risk framework uses a version of this idea, whether it is called ATR sizing, volatility targeting, or risk parity.

What still applies and what does not

Element	Still valid?	Comment
Volatility-based unit sizing	Yes, entirely	The single most transferable idea in the system
Diversification across uncorrelated markets	Yes	Still the mechanism that makes trend following tolerable
Pyramiding with a shared stop	Yes	A sound anti-martingale structure
Drawdown-based size reduction	Yes	Extends survival; used widely today
Specific 20 and 55 day parameters	Treat as illustrative	Adopting exact 1983 parameters is copying a fit, not a principle
Taking every signal	Yes, in spirit	Skipping signals removes the rare large winners
Pure breakout entries	Weaker than before	More competition; many implementations add volatility or trend filters
Futures-only universe	Optional	ETFs and micro contracts make the structure accessible to smaller accounts

The behavioural result, which was the point

The most cited detail of the experiment is that the Turtles received identical rules and produced very different results. Some followed the system precisely; others skipped signals after losing streaks, hesitated on entries, or reduced size at the wrong moments. The rules were necessary but not sufficient.

This is the strongest available evidence for a claim made throughout this library: execution consistency matters at least as much as strategy selection. A mediocre system followed exactly will usually outperform an excellent system followed selectively, because selective following removes precisely the uncomfortable trades that generate the returns. See trading psychology.

Frequently asked questions

Do the Turtle rules still work?

The architecture does; the specific parameters are dated. Pure Donchian breakout systems have seen returns compress as more capital pursues trend following and as markets have become more efficient. Modern implementations retain the volatility sizing and portfolio limits while using different entry filters and longer or blended lookbacks.

What is N in the Turtle system?

N is a 20-day exponential average of the true range, essentially the ATR. It measures a market’s typical daily movement, and the Turtles used it to convert every market into comparable risk units, so that a position in bonds and a position in coffee contributed similar volatility to the portfolio.

Can I trade the Turtle system with a small account?

Not in its original form, which requires enough capital to hold positions across many futures markets simultaneously. Micro futures contracts and liquid ETFs make a scaled-down version feasible from tens of thousands of dollars, with fewer markets and correspondingly lumpier results.

Why did the Turtles use two systems?

System 1 skipped breakouts that followed a winning breakout, which improved efficiency but risked missing a major trend. System 2, on a longer 55-day lookback with no skip rule, guaranteed participation in any large move. Running both meant the portfolio could not be left out of a historic trend by a filtering rule.

What is the most important thing to take from the Turtles?

That a complete system specifies sizing, pyramiding, correlation limits, and drawdown response, not just entries and exits. Most retail strategies specify only the parts that are fun to think about, and fail on the parts the Turtles treated as the core of the method.

Test this idea before you trade it

Describe the rules in plain language and AlgoTrader AI turns them into a structured strategy blueprint with a configurable historical backtest, cost assumptions, and exportable code.

Build a backtest

Keep reading

Referenced by

Educational use only. This guide explains how a strategy works. It is not investment advice, not a recommendation, and no result described here is a forecast. Test any approach on historical and out-of-sample data, size positions conservatively, and never risk money you cannot afford to lose.