Turtle Trading — The Legendary Donchian Channel Breakout System

In 1983, two Chicago traders — Richard Dennis and William Eckhardt — got into an argument over whether trading could be taught, or whether you had to be born with it. Dennis claimed he could train profitable traders "the way you raise turtles in Singapore." He placed an ad in a newspaper, picked a dozen or so people with no experience — a teacher, an accountant, a card player — taught them a simple, mechanical system, and handed them his own money. The group went down in legend as the Turtles, and by the participants' own accounts, they generated tens of millions of dollars in profit over the following years.

But the most interesting part of this story isn't that the experiment worked. The most interesting part is what separated the Turtles who made money from the ones who washed out — because everyone got identical rules. But let's take it in order.

What the Turtle Strategy Involves

The core of the system is the Donchian channel — a strikingly simple indicator: the upper band is the highest high of the last N candles, the lower band is the lowest low, and the middle is their average. No moving averages, no statistics — pure price. The logic: a market making a new N-period high is demonstrating strength, and big trends, by definition, have to make a series of new highs along the way.

The Turtles traded two variants:

• System 1 (shorter): a long entry on a breakout of the 20-day high (short: the 20-day low); exit on a break of the 10-day extreme in the opposite direction.
• System 2 (longer): entry on a breakout of the 55-day extreme; exit on a break of the 20-day one. Fewer signals, bigger trends — hence the popular shorthand "Donchian 20/55."

The other half of the system — in the Turtles' own view, the more important half — was risk management:

1. The N unit. Market volatility measured by ATR (Dennis called it "N"). Position size calculated so that a 1N move changes the account by ~1%.
2. A 2N stop. A hard stop loss set at a distance of 2×ATR from entry — roughly 2% of account risk per trade.
3. Pyramiding. Adding to a winning position every 0.5N of favorable movement, up to a limit of 4 units.
4. Portfolio limits. A maximum number of units per market and per group of correlated markets — so five positions don't turn out to be one and the same bet.

The original System 1 had one more, often-overlooked detail: a filter on the previous breakout. The 20-day signal was skipped if the previous breakout on that market had been profitable — the logic being that after a successful trend, a market usually needs to rest, and the next breakout more often fails. System 2 (the 55-day one) was played every time, as insurance against missing a truly huge trend. This detail says a lot about the system's character: even the rules for "when NOT to play" were mechanical, written down in advance, and non-negotiable mid-session.

Notice what's missing here: forecasting. The system doesn't predict where the market will go. It enters on every strength signal, cuts false breakouts quickly, and lets the few that turn into real trends run. It's a pure breakout strategy fused with iron risk control — trend following in its purest form.

What the Numbers Say

Historically: Dennis started with a few hundred dollars and grew it into tens of millions, and the Turtles as a group — by their own accounts — made tens of percent a year in their best years. Treat those numbers as founding legend, not benchmark: 1980s futures markets were far less efficient, and the results come from participants' own stories, not an audit.

The more interesting question is whether the mechanics work today, and in crypto. According to tests of the Donchian 20/55 variant on BTC daily data since 2017, the strategy was profitable through both the bull and the bear market, with the profile typical of trend following: a 30–40% win rate, but an average win 3–5 times larger than the average loss (data from published backtest summaries, not independently verified by us — reproduce the test yourself before risking real money). Intuitively it fits: BTC is a market that has historically behaved in a momentum-driven way and regularly funded trends lasting many months — exactly the environment this system was designed for.

Let's pause on that win rate, because that's where the substance lies. A 35% win rate with wins 4× the size of losses gives a solidly positive expectancy: over 100 trades, 35 winners × 4R minus 65 losers × 1R = +75R. The math is on your side. But that same math also says something else: at a 35% win rate, the probability of eight losses in a row somewhere in a run of 100 trades is over 90%. Streaks of 5–6 losses will happen repeatedly every year. That's not a system malfunction — it's its statistical everyday reality.

And that's the real answer to why some Turtles washed out despite identical rules: the rules didn't differ, but the ability to execute them through a losing streak did. Eckhardt won his bet only halfway — trading can be taught, but discipline is a much harder sell.

How to Apply It Step by Step

A modern, simplified version of the 20/55 system for crypto (D1):

1. Market: liquid assets with a history of trends — BTC, ETH. Thin altcoins churn out false breakouts by the dozen.
2. Long entry: a daily candle closes above the 55-day high. (Shorter variant: the 20-day high — more signals, more false ones.)
3. Initial stop: 2×ATR(20) below the entry price.
4. Position size: from the %-risk model — 1% account risk to the 2N stop.
5. Management: exit when price closes below the 20-day low (a natural structure-based trailing stop). No take-profits — the trend should say for itself when it's over.
6. Signal discipline: you take every signal. Hand-picking the "better" signals wrecks the system, because the whole year's result comes from 2–4 trades — and you don't know in advance which ones.

Numerical example on BTC (illustrative): a $10,000 account, 1% risk = $100. BTC consolidates for months under $90,000; the 55-day high is $90,500. The daily candle closes at $91,200 — a signal. ATR(20) = $2,800, stop = 91,200 − 5,600 = $85,600. Position = 100 / 5,600 ≈ 0.018 BTC. Scenario A: a false breakout, the stop gets taken out → −$100. Scenario B: the market launches into a trend; after three months, price is $128,000, the 20-day low has risen to $118,000, and that's where the exit lands → a profit of about $26,800 per BTC ≈ +$480 (4.8R). There will be a handful of scenario-B trades a year; scenario-A ones — a dozen or more. The annual result is decided by that handful.

A practical piece of advice before starting: before you put down a single dollar, run the rules above against 3–5 years of BTC daily data in any backtesting tool and work out three things — the win rate, the average-win-to-average-loss ratio, and the longest losing streak. Not to "optimize the parameters" (leave 20/55 alone), but to see with your own eyes what a year of this system looks like from the inside. Anyone who knows their historical worst losing streak has a reference point during a live one instead of panic.

When It Doesn't Work and Common Pitfalls

• A sideways market. Enemy number one. In a months-long consolidation, every signal is a false breakout and a loss — the system can post a year of "sawing" its way down by double digits. Filters (e.g., a wider trend on W1, rising volume on the breakout) reduce the problem but don't eliminate it.
• The psychology of losing streaks. Worth repeating: at a 30–40% win rate, long losing streaks are as certain as taxes. Anyone who "fixes" the system or skips a signal after the fifth loss in a row has just missed the trade meant to pay for the whole year. If you can't stomach 8 losses in a row, this isn't the system for you — and it's better to know that before you start.
• Thin markets. On small-cap alts, breakouts can be painted by a single aggressive candle and reversed immediately. The system was designed for liquid futures markets; porting it onto memecoins is a caricature of the original.
• A stop too tight "to improve the RR." A 0.5N stop instead of 2N looks better on paper, but it shakes you out on the market's first breath — and the system lives on letting trends breathe.
• Pyramiding without limits. Adding to a winning position accelerates gains and losses alike. The original limits (max 4 units, correlation limits) existed for a reason — without them, one sharp reversal wipes out a quarter.
• Believing "published means it works forever." The turtle rules have been public for decades; on classic futures markets, their edge has weakened. Crypto looks friendlier to this mechanic for now, but that's a claim with an expiry date — the only honest answer is your own, regularly refreshed backtest.

The no-hype conclusion: turtle trading is one of the few systems in trading history with complete, public rules and a documented origin story — and that's exactly why it's worth knowing, even if you never play a single one of its signals. It teaches three things that carry over to every style: that an edge can exist at a 35% win rate, that position size and the stop matter more than the entry signal, and that the weakest link in any mechanical system is the human being who has to execute it. Dennis proved that anyone can be taught the rules. The market has spent forty years proving that discipline is a different story.

FAQ

Rafał — Krypto Bez Ściemy

Trader and founder of Strefa Tradingu. He’s been breaking crypto down on YouTube for years — no hype, no signals, with a focus on market structure and risk management.

kryptobezsciemy.plInstagramYouTubeKlub Zero