In 1983 Richard Dennis bet a colleague that trading could be taught to anyone. He recruited a group with no market background, gave them a set of rules over two weeks, and funded them. Several went on to make substantial money. The core entry rule was as simple as rules get: buy when price makes a new high of the last twenty days.
Every riverside town has a mark on a wall showing the highest the water reached. Nobody needs a model to read it. When the water passes that mark, something is happening that has not happened in living memory, and everyone reacts.
A twenty-day high is that mark. It requires no interpretation, no parameters to argue about, and every participant can see it. That visibility is part of why it works — the level matters because people act on it.
What a Donchian channel is
- N = 20
- The classic entry channel — roughly a month of sessions
- N = 55
- The slower channel the Turtles used for a second system
- N = 10
- A common exit channel, faster than the entry
Example: A stock whose highest high over twenty sessions was ₹482 has an upper channel at ₹482. A close above it is a breakout by definition — no judgement required.
The part everybody skips
The entry rule is what gets quoted. It is also the least important part of the system, and quoting it alone is how the Turtle rules acquired a reputation for not working.
- 1Position size from volatility, not from conviction
Every position was sized so that a move of one average true range meant the same rupee amount, whatever the stock. A volatile name got fewer shares. This is where most of the risk control lived.
- 2A stop that was set before entry
Two average true ranges below entry, decided at the moment of buying and never widened. The rule was not "buy breakouts"; it was "buy breakouts and accept being wrong at a pre-set price".
- 3An exit channel faster than the entry
Enter on a 20-day high, exit on a 10-day low. Exiting faster than you enter is what keeps a winner from giving everything back — and it feels wrong every time.
- 4Take every signal
The returns came from a small number of very large trends. Skipping signals because one looked unconvincing was the fastest way to miss the trades that paid for everything else.
Where it struggles in India
- Range-bound markets punish it. In a sideways year the rule buys every false high and sells every false low, bleeding steadily. An ADX filter — only take breakouts when ADX is above 25 — removes a lot of that, at the cost of missing some genuine starts.
- Circuits break the exit. A stock hitting the lower circuit gives you a breakdown you cannot act on. The rule assumes you can always trade at the level; in Indian small caps that assumption fails exactly when it matters.
- Gaps skip the stop. Overnight news gaps straight through a stop level. The pre-set stop is still the right discipline, but the realised loss can be considerably larger than planned.
- Costs compound. A system taking many signals across many names pays brokerage, STT and slippage on all of them. On a mid-cap with a wide spread, that drag alone can consume the edge.
The chart workbench draws the 20-day Donchian channel as a band overlay, and the rule tester runs the simplified version — buy a 20-day high, exit a 20-day low — against whatever stock and range you have loaded, charging realistic costs. Run it across several stocks and several windows before drawing any conclusion from one result.
You test the 20-day breakout rule on a stock and find it wins on only 31% of trades. What should you conclude?
Nadi kinare har gaon mein deewar pe nishaan hota hai — paani sabse upar yahan tak aaya tha. Koi calculation nahi chahiye, sabko dikhta hai. Bees din ka high wahi nishaan hai. Par asli kaam entry nahi thi — size aur stop pehle se tay karna tha.
- A Donchian channel is the highest high and lowest low of the last N bars — nothing to interpret.
- The published entry rule is the least important part of the Turtle system.
- Volatility-based sizing and a pre-set stop carried most of the risk control.
- Exiting on a faster channel than you entered is what protects a winner.
- Trend systems win a minority of trades; judging them by hit rate guarantees abandoning them.
Mark it done to track your progress through the curriculum.
Common questions
Short, direct answers to what people ask about this topic.
- donchian channel meaning
- A Donchian channel plots the highest high and the lowest low of the last N bars as an upper and a lower band, with the midpoint between them as a middle line. Nothing in it is interpreted or drawn by hand, so a close above the upper band is a breakout by definition. That mechanical quality is its main advantage over chart patterns, which you can talk yourself into seeing when you want one.
- the turtle traders entered a long position when price made a new
- Twenty-day high — the classic Donchian entry channel, roughly a month of sessions. In 1983 Richard Dennis bet that trading could be taught to anyone, recruited a group with no market background, gave them the rules over two weeks and funded them, and several went on to make substantial money. That entry rule is the part everyone quotes and the least important part of the system.
- what are the standard donchian channel periods
- Twenty bars for the entry channel, ten for the exit channel, and fifty-five for the slower second system the Turtles ran alongside it. Exiting on a faster channel than you entered — in on a 20-day high, out on a 10-day low — is what stops a winner giving everything back, and it feels wrong every single time.
- why does a donchian breakout system lose most of its trades
- Because most breakouts fail, and a trend-following rule is built to accept that. Such a system typically wins on a third of its trades or fewer and makes money only because the winners run several times further than the losers. Hit rate on its own says almost nothing: 31% winners with an average winner four times the average loser is a very different result from 31% winners with equal-sized outcomes.
- what makes the turtle breakout rule harder to run on indian stocks
- Three frictions the original rules never had to price in. Circuit limits hand you a breakdown you cannot act on when a small cap locks at the lower circuit; overnight gaps skip straight through a pre-set stop and make the realised loss larger than planned; and brokerage, STT and slippage are paid on every one of a large number of signals. Range-bound years punish it separately, since the rule buys every false high and sells every false low.