Turtle Trading: The Original Rules of Richard Dennis

In 1983, a commodity trader from Chicago bet that trading can be learned like a craft. Richard Dennis recruited people with little or no market experience through newspaper ads, trained 13 of them for two weeks and gave them real money. Over the following four years the group earned an average of 80% a year. The rulebook behind it is still the best-known mechanical trading system in market history.

This guide spells out the key rules instead of just mentioning them. It is based on the original document “The Original Turtle Trading Rules”, which former Turtle Curtis M. Faith and other participants published for free in 2003. You will find the entry signals, the position size calculation, the stop rule and the exits here, each with the numbers from the original. This article follows our editorial policy.

Turtle trading in 30 seconds

Turtle trading is a rule-based trend-following system that enters on a breakout from a 20-day or 55-day price channel and calculates position size from the average daily range. Entries, exits and position sizes follow fixed rules. All values below come from the original document of 2003:

• Two entry systems: System 1 buys a breakout above the 20-day high, System 2 above the 55-day high. Short trades work the same way below the low.
• The volatility unit N: position size depends on a 20-day average of the true range, today known as the ATR. Dennis and Eckhardt called this value N.
• One unit: a position unit is sized so that a move of 1 N equals 1% of account equity.
• Stop at 2N: each unit could risk at most 2% of the account, so its stop was 2 N from the entry. A fully built position can lose more in total, see below.
• Adding in half-N steps: if the trend runs, more units are added every 0.5 N, at most four per market.
• Exit against the trend: System 1 exits at the 10-day low, System 2 at the 20-day low.
• Position limits: at most 4 units per market, 6 in closely correlated markets, 10 in loosely correlated markets and 12 per direction.
• The documented result: Faith puts the group’s return over four years at an average of 80% a year.

What is Turtle trading?

Turtle trading is a rule-based trend-following system in which a position is opened on a breakout from a 20-day or 55-day price channel and sized by the average daily range of the market.

The name goes back to a remark by Richard Dennis. He had seen turtle farms in Singapore and said he would grow traders the same way. From then on the participants were called Turtles.

The group did not invent the principle. Its core, entering on a breakout from a consolidation and adding only to winners, can already be found in the trading rules of Jesse Livermore from the 1920s. Dennis turned it into a mechanical system that works without personal judgment. How breakouts work in general is in our guide to breakout trading.

It is important to separate the original from what is sold under this name today. The original is a pure futures system for 20 to 25 markets. It was never meant for stocks, crypto or intraday trading. Whoever transfers it to other markets is trading a variation, not the Turtle rules.

The Turtle trading rules at a glance

Faith describes the system as a complete trading system. By that he means it answers every question a trader has to ask: what, when, how much and where to get out again. The table summarizes the building blocks with the original values, the details follow below.

Building block	Rule in the original	What it means
Markets	20 to 25 liquid futures markets	spread across commodities, rates, currencies
Position size	1 unit = 1% of account per 1 N	volatile markets get smaller positions
Entry	20-day or 55-day breakout	two systems, can be combined
Adding	every 0.5 N, at most 4 units	add only to winners
Stop	2 N from entry, 2% risk	fixed, no exceptions
Exit	10-day low (S1), 20-day low (S2)	let profits run until the reversal

The markets

The Turtles traded only liquid futures contracts. These included interest rate products, currencies, commodities such as heating oil, crude oil, gold and silver, and stock indices.

Broad diversification is not an extra but part of the math. Faith writes that most of the profits in a given year might come from only two or three large winning trades. Whoever watches only a few markets misses exactly these outliers and keeps the many small losses.

Position size: the volatility unit N

N is the most important term of the whole system. Faith defines it as the 20-day exponential moving average of the true range, the actual daily range including price gaps. Today the same idea is known as the average true range, ATR for short.

True range = the largest of (high − low), (high − previous close), (previous close − low)

N = (19 × previous day’s N + today’s true range) / 20

Position size follows from N. First you calculate the dollar volatility: N times the dollar value of one point in the contract. Then comes the position unit, which the Turtles called a unit.

Unit = 1% of account equity / (N × dollars per point)

The point is to normalize across all markets. One unit of gold and one unit of heating oil move the account by the same amount on an average day, although the two markets fluctuate completely differently. Only this makes diversification across many markets comparable. How to size positions in general is in our guide to position sizing.

The entry: System 1 and System 2

The Turtles got two entry systems and were free to choose. Both are based on the channel breakout of Richard Donchian. Some participants traded only System 2, others split their capital half and half.

• System 1: enter as soon as price exceeds the high of the last 20 days by a single tick. Go short when price falls below the 20-day low.
• System 2: enter when price exceeds the high of the last 55 days, short at the 55-day low.

One detail decides whether it can be traded at all. The Turtles traded the breakout immediately during the session and did not wait for the close. If a market opened with a gap beyond the level, they entered at the open.

The filter that makes System 1 work

This rule is missing from almost every description of the system, and without it System 1 does not work. A 20-day breakout was skipped if the previous breakout in the same market would have been a winner. So the Turtles only traded after a failed signal.

A breakout counted as a loss if price then moved 2 N against the position before the regular exit at the 10-day low. The direction did not matter: a failed short allowed the next long.

So that the Turtles did not miss big moves, there was a safety net. If an entry was skipped because of this filter, they entered at the latest on the 55-day breakout. Faith calls this the failsafe breakout.

Linda Bradford Raschke later turned this logic around. Her setup Turtle Soup trades exactly the failed 20-day breakouts on which the Turtles lost money.

Adding in half-N steps

The Turtles entered with a single unit and built the position only in profit. Each further unit came after a move of 0.5 N, measured from the actual fill price of the previous order. At most four units per market were allowed.

Step	Price in the original gold example	Distance
First unit	310.00	55-day breakout
Second unit	311.25	+ 0.5 N
Third unit	312.50	+ 0.5 N
Fourth unit	313.75	+ 0.5 N

In fast markets, all four units could come together on a single day. That is the core of the system: in a strong trend the full position is on, in a sideways phase it stays at one unit. In the example, N is 2.50.

The stop at 2N

This is the number most often reported wrong. Faith writes that no trade could carry more than 2% risk. Because a price move of 1 N equals 1% of the account, the stop is 2 N below the entry, or 2 N above it for short positions.

When more units were added, the old stops moved up. Each time, they were raised by 0.5 N, so that in the end all stops sat at the same price, 2 N from the most recently added unit. This limits the risk of the total position instead of letting it grow with every unit.

There was a documented alternative called the Whipsaw. Here the stop was only 0.5 N, so 0.5% risk. Faith writes that this variant was more profitable but harder to stick with, because it produces many more losing trades. More on the logic behind it is in our guide to risk management.

Another rule applied at account level. If a Turtle was down 10%, they kept trading as if their account were only 80% of its size. After another 10% loss, the size was cut by a fifth again. The system shrinks itself when it is not working.

The exit: the hardest rule of the system

The Turtles did not exit at a price target but on the opposite breakout. System 1 closes a long position at the 10-day low, System 2 at the 20-day low. For shorts, the corresponding highs apply. There is no price target anywhere in the rulebook.

Faith himself calls this the hardest part of the system. He writes that waiting for a new 10-day or 20-day low often means watching 20%, 40% or even 100% of a significant paper profit evaporate. That is where most people who rebuild the system fail.

The position limits

Four limits capped the total risk across all markets. They prevent several positions from adding up to one big bet.

Level	Maximum	Example from the original
Single market	4 units	4 units of Japanese yen
Closely correlated	6 units per direction	heating oil and crude oil, gold and silver
Loosely correlated	10 units per direction	gold and copper
One direction	12 units	12 long and 12 short possible at the same time

For a full position in a market, the Turtles used the word “loaded”. Four units of yen meant fully loaded in yen. If you add up the numbers, you see the real burden. In the gold example above, all four stops sit at 308.75, 2 N below the last unit. A stop-out would cost 1.25 + 2.50 + 3.75 + 5.00 points, or about 5% of the account measured from the entry prices. Price gaps can make the loss larger.

A worked example from the original document

Faith calculates the unit size for heating oil. I use his example unchanged, because it shows how small one unit is even with an account of one million dollars.

Item	Value
Market	heating oil, March 2003 contract
N on December 4, 2002	0.0141
Account size	$1,000,000
Dollars per point	$42,000 (42,000 gallons)
Calculation	(0.01 × $1,000,000) / (0.0141 × 42,000)
Result	16.88, rounded down to 16 contracts

Two things stand out in this calculation. First, it is rounded down, because you cannot trade fractions of contracts. Second, the Turtles received the unit sizes once a week on a sheet of paper, every Monday. Nobody recalculated daily.

For a private account, the finding is uncomfortable. With $50,000 the same calculation would land below one contract. The original system needs either a lot of capital or smaller contract sizes, such as micro futures; otherwise position size cannot be applied cleanly.

The experiment: Richard Dennis and the bet

In 1983, two friends argued about an old question. Richard Dennis was convinced he could teach trading. His partner William Eckhardt believed talent was decisive. To settle it, they wanted to train beginners and give them real accounts.

The ad ran in Barron’s, the Wall Street Journal and the New York Times. More than 1,000 applications came in, and Dennis held 80 interviews. Ten candidates remained, and he added three people he knew. These 13 people formed the first class and were trained for two weeks at the end of December 1983.

From January they traded small accounts, from February real ones. Faith gives sizes between $500,000 and $2,000,000 for most participants. A second class followed in 1984, which is why many accounts give a total of 23 Turtles.

The original document gives exactly one number for the result. Over the following four years, the group earned an average annual compound return of 80%. The often quoted “more than $100 million profit” is not in it; such figures come from later books and cannot be verified from the document.

Dennis drew a sober conclusion. The Wall Street Journal quoted him in 1989 as saying that trading was even more teachable than he had imagined, and that this was almost humbling in a strange way. Why he later lost tens of millions himself is in his portrait linked at the top of this guide.

Does Turtle trading still work today?

The honest answer is: not unchanged. Faith wrote in 2003 in his foreword that most former Turtles were by then trading even better rules. He also wrote that he considered it unlikely that publishing the rules would teach many people to trade like the Turtles.

He gives three reasons. First, the rules sold by others would not be clear, because the sellers did not know how to trade. Second, and more important, even with clear rules most buyers would not be able to follow them. Third, most Turtles were by then trading better rules anyway. Whoever sits out after the third false breakout in a row is trading a different system.

What still works about the Turtle system

• Volatility-based position size: sizing positions by volatility is standard in almost every professional approach today.
• Fixed stop before the entry: the exit price is set before the position exists. That takes the decision out of the running trade.
• Adding only to winners: adding to losers is the most common way to ruin an account. The system rules it out by design.
• Fully testable: every rule is unambiguous, so the system can be tested instead of debated.

Where the original fails today

• High capital requirement: the unit math needs an account large enough to size at least one contract per unit in a diversified set of markets. With full-size contracts, five-figure capital rarely gets there.
• Long losing streaks: trend following produces many small losses and a few large gains. Faith describes phases in which much of the paper profit is given back.
• Known levels: 20-day and 55-day breakouts have been public for more than 20 years. The original document does not show whether these rules still have an edge today after costs.
• No intraday use: the system works on daily data. It is not made for short-term approaches.

Whoever wants to test the rules needs daily data and patience. A clean test includes the System 1 filter, otherwise it measures a different system. How to run such a test yourself is in our guide to backtesting.

Curtis Faith: Way of the Turtle

Curtis Faith was the youngest participant of the experiment in 1983 and later wrote the most detailed book about it, “Way of the Turtle”. I have read it and place it here, because people often look for the book together with the rules.

The book is not a rulebook but an account of experience. If you are looking for the pure trading rules, you will find them more compactly in the original document. Faith’s strength lies elsewhere: he describes why participants with identical rules got completely different results.

The most instructive part is about those who deviated. The rules document says it plainly: the Turtles with the best results applied the entry rules consistently, while those with the worst results, and all who were dropped from the program, had failed to take signals. The book explains why that happened. That is the real finding of the experiment, more important than any single number in the rulebook.

A second part deals with the trading edge. Faith argues that an edge does not come from a secret signal but from the expected value over many trades.

My assessment after reading it: worth it for anyone who wants to understand why discipline weighs more than signal quality. If you expect a step-by-step guide, you will be disappointed.

Where to find the original Turtle rules

The complete rulebook is freely available. Curtis Faith and other former Turtles published it in 2003 as “The Original Turtle Trading Rules”, explicitly for free, after a former participant had sold the rules for a lot of money. The authors keep it on the official site of the Original Turtles. The document has 38 pages.

Richard Dennis took part in the discussions that led to the free release. According to the document, he and William Eckhardt were not consulted before a former Turtle marketed the rules and did not profit from that sale. The Turtles’ contractual confidentiality agreement ended in late 1993.

Conclusion: the rulebook matters more than the numbers

The Turtle trading system is no secret any more, but a lesson. Its value today lies less in the specific levels 20 and 55 than in its design: position size from volatility, a fixed stop before the entry, adding only to winners, exits by rule instead of by feeling.

Whoever wants to adopt the rules unchanged needs enough capital for a diversified futures portfolio and the nerves to give back large paper profits. Few manage that, and Faith says so clearly in his own document. The realistic way is to take individual building blocks into your own approach.

The real finding of the experiment is not about the rules. Dennis proved that trading can be learned. Those who failed were the participants who deviated from the rules, not those who lacked talent.

Frequently asked questions about Turtle trading

How much did a Turtle risk per trade?

At most 2% of account equity per unit. A price move of 1 N equals 1% of the account, and each unit’s stop was 2 N from its entry. A full position of several units could lose more in total, and price gaps could make any loss larger. The often quoted figure of 1% per trade is not what the original document says.

What is the difference between System 1 and System 2?

The length of the price channel. System 1 enters on a breakout above the 20-day high and exits at the 10-day low, System 2 uses the 55-day high and exits at the 20-day low. System 1 also has a filter: it skips a breakout if the previous one would have been a winner.

What does N mean for the Turtles?

N is the average daily range of a market. It is the 20-day average of the true range, today known as the average true range or ATR. Both the position size and the stop follow from N.

How many Turtles were there?

The first class at the end of 1983 had 13 people. Dennis chose 10 from more than 1,000 applications and added three people he knew. A second class followed in 1984, which is why a total of 23 is often given.

How successful was the experiment really?

The original document gives 80% a year over four years. Absolute profit figures such as “more than $100 million” come from later books and are not in the Turtles’ own rulebook.

Can the system be used for stocks or crypto?

Only with compromises. The original is a pure futures system for 20 to 25 markets. A transfer is technically possible but changes the foundations, because contract values, trading hours and liquidity are different. The result is a variation, not the Turtle system.

Where can I get the original Turtle rules?

For free, on the official site of the Original Turtles. The document “The Original Turtle Trading Rules” by Curtis M. Faith and other former Turtles from 2003 has 38 pages.