Donchian Channel Strategy: Breakout Rules, Turtle Differences and Testing
A Donchian breakout strategy compares price with a range established by earlier bars. Its central implementation requirement is to exclude the signal bar from the breakout threshold. A close cannot exceed the highest high of a window that includes that same close's bar.
How the Channel Works
The upper line is the highest high over N bars; the lower line is the lowest low; the middle is their midpoint. The middle line is not the average closing price. TradingView's Donchian Channels documentation describes these calculations and the length and offset controls.
Distinguish the displayed channel, which can include the developing bar, from the threshold used to trade. For a daily close breakout, compare today's completed close with the highest high of the previous 20 completed daily bars. Compute the short threshold from their lowest low. A visual plot offset alone does not change the data used by a strategy condition.
A new high establishes a historical observation. It does not establish that a profitable trend will follow.
A Reproducible Breakout Baseline
The following is a proposed research specification, not a tested performance result:
| Decision | Long-only daily baseline |
|---|---|
| Entry | Completed close exceeds the previous 20 bars' highest high while flat |
| Order timing | Submit after confirmation; model the next available fill |
| Initial stop | Fill price minus twice the ATR recorded at the signal |
| Trend exit | Completed close below the previous 10 bars' lowest low |
| Position size | Cash risk budget divided by entry-to-stop distance, rounded down |
| Re-entry | A new qualifying event after the prior position has closed |
Define what happens when a protective stop and a close-based exit occur on the same bar. Record commissions, spread, slippage and overnight gaps. A stop is an execution instruction; a gap can produce a larger loss than the intended risk budget. Futures also require contract point value and margin constraints.
Waiting for a close can remove some intrabar breakouts that reverse before confirmation. It also delays entry and changes the price paid. There is no documented percentage reduction in false signals for this baseline.
What the Turtle Rules Actually Specified
The original Turtle rules, available in this copy of the original document, describe intraday breakouts of preceding ranges. They did not generally wait for the daily close.
| System | Entry range | Opposite breakout exit |
|---|---|---|
| System 1 | Previous 20 days | Previous 10 days |
| System 2 | Previous 55 days | Previous 20 days |
System 1's skip rule evaluated whether the previous breakout would have won, including a breakout that had been skipped. A 55-day failsafe addressed a missed trend. This is more specific than “skip after your last winning trade.” System 2 did not use that filter.
The document also specifies volatility-based units, additions, stops and portfolio limits. Quoting a 2% risk rule without those details does not reproduce the system. The simple close-based baseline above is a separate experiment.
Pullbacks, Stops and Channel Width
A midpoint pullback entry is another strategy. Define a touch, a rejection and the time allowed for confirmation before testing it. Merely touching the midpoint does not show buying support. Compare its missed trends and entry prices with the breakout baseline under identical costs.
For a trailing stop, decide whether the lower channel can move downward. A rolling low can fall when new lows enter the window; a protective long stop intended only to tighten should retain the higher of its previous value and the new candidate. A close-based exit and an intraday stop order have different fill assumptions.
Channel width measures the distance between window extremes. It can narrow when an old extreme leaves the window, without a new change in trading activity. It can widen after an isolated gap. Neither condition proves that a breakout is imminent or that a trend will continue.
Testing Confirmation Filters
A moving-average slope, RSI threshold or volume requirement is a hypothesis to test separately. All can reject profitable trades as well as losing ones. An RSI above 70 is not proof of exhaustion, and broker tick volume is not consolidated traded volume.
Keep a ledger with the signal timestamp, prior channel levels, fill, stop, exit reason and costs. Compare an unfiltered baseline with each filtered variant on the same symbols and dates. Reserve later data before selecting parameters. Report trade count, average win and loss, drawdown, exposure and return after costs; win rate alone cannot establish an advantage.
For equities, document splits and dividends. For futures, specify contract rolls and tradable contract prices. For currency pairs, include financing and session conventions. These choices can materially change a breakout test.
Getting Started in TradingView
- Add Donchian Channels to a standard candlestick chart and record the symbol, session and timeframe.
- Set a trial length such as 20 and inspect several completed bars against their preceding ranges.
- Write the entry, stop, exit and sizing rules before creating a strategy.
- Verify the strategy's trade list against chart events, including gaps and bars touching both entry and stop levels.
- Forward-test the frozen rules and record differences between simulated and actual fills.
Pineify can help draft a Pine strategy from explicit rules. Review its indexing and order timing, request syntax checking of the final version, and inspect behavior in TradingView. Generated code does not establish a strategy's profitability.
Frequently Asked Questions
Is 20 the best channel length? No universal optimum is established here. A shorter range changes the number and timing of breakouts. Compare a small prespecified set on data excluded from parameter selection.
Can Donchian Channels be used intraday? The calculation can use intraday bars. Costs, session boundaries and execution assumptions then become especially important.
What win rate should I expect? No verified backtest accompanies this article. A return estimate requires a complete strategy, data and cost model; the channel alone has no win rate.
Does discipline make the rules profitable? Consistent execution makes a test interpretable. It cannot create an advantage in rules that lose after costs.
