Turtle Trading Strategy: Breakout Rules, ATR Position Sizing and Why Choppy Markets Hurt

Most traders who lose money on breakouts don’t lose because the idea is wrong. They lose because they buy the breakout with no fixed size, no fixed stop and no plan for the dull months when nothing trends. The turtle trading strategy is interesting precisely because it answered all three questions in writing, back in the 1980s, and then handed the answers to people who had never traded before.

This piece walks through the actual rules: the 20-day and 55-day breakouts, the volatility measure the Turtles called N, how a position “unit” was sized, where the stops went, and how they exited. Then we run the numbers on a hypothetical NSE stock in rupees and look honestly at why the system spends long stretches losing money. The rule details come from The Original Turtle Trading Rules (originalturtles.org, PDF), and the history from Investopedia’s account of the Turtle experiment.

Where the turtle trading strategy came from

The story starts with a bet. Richard Dennis, a Chicago commodities trader, believed trading could be taught. His friend and fellow trader William Eckhardt thought Dennis had a gift that couldn’t be passed on. To settle it, Dennis advertised in The Wall Street Journal, picked a group of novices, trained them for two weeks and then gave them his own money to trade, according to Investopedia. He called them “turtles” after turtle farms he had seen in Singapore.

Wikipedia’s page on Richard Dennis says he trained 21 men and two women in two groups, one from December 1983 and the other from December 1984. After the trial period, those who applied the system properly were given accounts of $250,000 to $2 million of Dennis’s money. Both Investopedia and Wikipedia put the group’s reported profit at $175 million over five years.

That headline number is why the system is famous. The less quoted part is what came later, and we’ll get to that.

The two breakout systems

The Turtles ran two related entry systems. Both are plain price breakouts, the same idea you see drawn on a Donchian channel.

• System 1 entered when price exceeded the high or low of the preceding 20 days by a single tick. Above the 20-day high, buy one unit. Below the 20-day low, sell one unit short.
• System 2 did the same with a 55-day breakout. It was the slower, longer-term version.

System 1 had a twist that many simplified versions leave out. A new 20-day breakout signal was ignored if the previous breakout in that market would have been a winning trade. The rules define a losing breakout as one where price moved 2N against the position before a profitable 10-day exit happened. If a System 1 signal was skipped this way, the trader still entered at the 55-day breakout, which the rules call the failsafe breakout, so a big trend wasn’t missed entirely.

The logic is a bet on clustering. After a winning breakout, the next one is more likely to fail, so the system waits for a stronger move. After a loser, it takes the next signal at the closer 20-day level.

N: the volatility number behind everything

Every size and every stop in the system is expressed in N. The rules define N as the 20-day exponential moving average of the true range, which today most people know as the average true range (ATR).

True range for a day is the largest of three values: today’s high minus today’s low, today’s high minus the previous close, and the previous close minus today’s low. The previous-close terms are there so overnight gaps count as part of the day’s movement.

N is then updated daily with this formula:

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

Because it needs yesterday’s N, you start with a simple 20-day average of true range and roll forward from there. Quick check with made-up numbers: if yesterday’s N was ₹40 and today’s true range is ₹60, the new N is (19 × 40 + 60) ÷ 20 = (760 + 60) ÷ 20 = ₹41. One wild day nudges N; it doesn’t jolt it.

Position sizing: what a “unit” means

This is the part of the turtle trading strategy that has aged best. Positions were built in pieces called units, and a unit was sized so that a 1N move in price equals 1% of account equity. The rules give it as:

Unit = 1% of account ÷ (N × money value per point)

For a cash stock, one point is one rupee per share, so the formula reduces to 1% of the account divided by N. A volatile stock gets fewer shares, a quiet one gets more, and each position tends to swing by roughly the same rupee amount on a normal day. The rules call this volatility normalisation very important, because trades in different markets end up carrying similar chances of a given rupee gain or loss. If this sounds like fixed fractional position sizing with a volatility adjustment bolted on, that’s exactly what it is.

The entry rule is the least special thing about the turtle system. A 20-day breakout is something anyone can draw. The sizing and the exits are what made it a system.

The rules also capped total exposure in units: a maximum of 4 units in a single market, 6 units in closely correlated markets, 10 units in loosely correlated markets, and 12 units in one direction overall, long or short.

Stops, adding units and exits

Stops. No trade could risk more than 2% of the account. Since 1N equals 1% of equity, the stop sat 2N below the entry for a long and 2N above for a short. The rules describe these as non-negotiable: if price hit the stop, the position was closed, every time.

Adding units. The Turtles entered one unit at the breakout and added more at ½N intervals as price moved in their favour, up to the 4-unit limit. Each time a unit was added, the stops on the earlier units were raised by ½N, so the whole position generally ended up with its stop 2N below the latest unit.

Exits. Winning positions were not closed at a target. System 1 exited longs on a 10-day low and shorts on a 10-day high. System 2 used a 20-day low or high. When price broke that exit level against the position, all units came off together.

The rules admit these exits are hard to live with. Waiting for a 10-day or 20-day low means watching a chunk of open profit disappear before you get out. That is the price of staying in long enough to catch the big trend.

A rupee worked example on an NSE stock

Everything below is hypothetical, picked to keep the maths clean. Say we have a ₹10,00,000 account and a stock trading near ₹1,000 with an N of ₹40.

• 1% of the account is ₹10,000.
• One unit = ₹10,000 ÷ ₹40 = 250 shares, worth about ₹2,50,000 at ₹1,000.
• The stock closes above its 20-day high and we buy 250 shares at ₹1,000.
• Stop = 2N = ₹80 below entry, so ₹920. If hit, the loss is 250 × ₹80 = ₹20,000, which is 2% of the account.

Now the stock keeps going. We add a unit every ½N, which is ₹20:

With all four units on, the common stop is ₹1,060 minus ₹80, which is ₹980. If the stock falls back to ₹980, the losses are ₹5,000 + ₹10,000 + ₹15,000 + ₹20,000 = ₹50,000, or 5% of the account. That is the worst case on a full position, and it’s worth knowing before you add the third and fourth units, not after.

Notice something else. Four units are 1,000 shares, worth roughly ₹10.3 lakh at an average price of ₹1,030. That is more than the whole ₹10 lakh account. The original Turtles traded futures, where margin made this possible. A cash-market trader in India can’t run the full 4-unit stack on one stock without leverage, so in practice you either cap yourself at two or three units, or accept that the system was designed for leveraged instruments and that leverage cuts both ways.

On the exit side, say the stock runs to ₹1,200 and then turns. With System 1 we sell everything when it breaks its 10-day low, wherever that is. If the 10-day low happens to be ₹1,150, we give back ₹50 a share from the top. Annoying, but the rule doesn’t care.

Shrinking size during a losing run

One rule gets skipped in almost every retelling. The Turtles were told to cut the notional size of their account by 20% each time they lost 10% of the original account. The rules’ own example: a $1,000,000 account down 10% would be traded as if it were $800,000, and a further 10% loss on that would cut it to a notional $640,000.

In our rupee version, if the ₹10 lakh account fell to ₹9 lakh, we’d size units as if we had ₹8 lakh. 1% becomes ₹8,000, and with N still at ₹40, a unit drops to 200 shares. It’s a built-in brake, and it matters more than it sounds, because of what comes next. It pairs naturally with the habits in our piece on risk management in trading.

Why the turtle trading strategy struggles in choppy markets

Trend following makes money in a handful of big moves and gives a lot of it back in between. Investopedia notes that drawdowns tend to be especially deep with trend-following systems, partly because most breakouts turn out to be false moves, and that practitioners say to expect to be right only 40% to 50% of the time.

Think about what a sideways market does to a 20-day breakout rule. The stock pokes above its 20-day high, we buy, it slips back inside the range and hits the 2N stop. A few weeks later it pokes below the 20-day low, we short, it bounces, another stop. Each loss is controlled, around 2% when only one unit is on, but they stack up. Wikipedia sums up the pattern plainly: this type of system loses money when the market is rangebound, often for months at a time, and profits during large moves.

A breakout system doesn’t fail in the trend. It fails in the months between trends, one small, perfectly sensible stop at a time.

There is also the question of whether the edge survived its own fame. Wikipedia says backtests of the published system show a drastic drop in performance after 1986 and even flat performance from 1996 to 2009. Investopedia’s bottom line is blunt too: the results are close to flipping a coin. So treat the rules as a framework for sizing and discipline rather than a money machine you can switch on.

Using the ideas on Indian stocks today

We think the parts worth borrowing are clear, even if the whole system isn’t:

• Size by volatility. Dividing a fixed rupee risk by ATR is one of the simplest upgrades a discretionary trader can make. It stops you from taking a giant position in a stock that routinely moves 5% a day.
• Know your stop before you enter. 2N is a starting point. The real lesson is that the stop is decided at entry and never moved further away.
• Write down the exit for winners. Most people have a stop for losers and nothing for winners. A trailing 10-day or 20-day low is crude, but it’s a rule.
• Cut size when you’re losing. The 10%-down, 20%-smaller rule is easy to copy and protects you when your timing is off.
• Respect costs and liquidity. A system that takes many small losing trades pays brokerage, taxes and slippage on every one. On thinly traded stocks, the breakout fill alone can eat the edge.

And test it on your own instruments before risking money. The rules themselves say every profitable system has a different optimal exit, and that you can’t judge the exit for winners without looking at how you handle losers.

The short version

The turtle trading strategy buys 20-day or 55-day breakouts, sizes each unit so 1N of movement equals 1% of the account, stops out at 2N, adds at ½N steps up to four units, and exits on a 10-day or 20-day reversal. It caught big trends in its day and paid for them with long, grinding drawdowns, and published backtests suggest the edge faded after the 1980s. The entry is ordinary. The sizing, the stops and the willingness to sit through bad months are the bits worth keeping.

Disclaimer: This article is for educational purposes only and is not investment advice. The stock, prices and account figures in the worked examples are hypothetical. Trading involves substantial risk, including the loss of capital. Please do your own research or consult a SEBI-registered adviser before making any investment decision.