Strategy guide · Price breakouts
Donchian Breakout Forex Strategy: Channels, ATR Sizing and Testing
A channel breakout turns a new trading-range extreme into a testable trend hypothesis. Define exactly which bars create the channel, when the signal becomes known and which exit controls the trade.
What a Donchian channel measures
A Donchian channel marks the highest high and lowest low over a chosen lookback. A breakout system treats movement beyond that range as a possible start or continuation of a trend. The channel is a record of past extremes; it does not predict how long a new move will last.
For a signal evaluated at today’s close, define the upper 55-day boundary as the maximum high of the previous 55 completed days, excluding today. Define the lower boundary similarly from their lows. Including today’s high makes a close above that same upper boundary impossible and is a common coding mistake.
The sample below uses closing confirmation and next-session execution. An intraday stop-entry version reacts sooner and has a different fill and false-breakout profile. Treat it as a separate system rather than mixing its best historical entries with the close-confirmed version.
Use ATR for distance, not direction
ATR measures range. True range is the largest of the current high minus low, the absolute high minus previous close, and the absolute low minus previous close. A moving average of these values gives ATR. MetaQuotes ATR documentation.
For this reference specification, use Wilder-style smoothing over 20 days: initialise with the mean of the first 20 true ranges, then calculate each new ATR as (19 × previous ATR + new true range) ÷ 20. Use the same calculation in research and execution; platform indicators can differ in defaults.
A two-ATR stop translates recent price variation into a repeatable distance. It does not mean a two-ATR adverse move is unlikely or impossible. If volatility rises sharply after entry, the pre-set stop remains a risk instruction, not a guarantee that the market will fill it at the chosen level.
A complete daily 55/20 rule set
- Use one fixed broker-time daily session definition and sufficient historical warm-up data.
- While flat, buy if the completed daily close is strictly above the prior 55-day high; sell if it is strictly below the prior 55-day low. Equality creates no signal.
- Execute at the first tradable quote in the next specified session window. Use ask for buys and bid for sells, including any gap.
- Set an initial stop two ATR(20) values from the actual entry, using ATR from the signal close. Allow one position per pair and no pyramiding in this baseline.
- For a long, a completed close below the prior 20-day low signals an exit at the next session’s tradable bid. For a short, a close above the prior 20-day high signals an exit at the next tradable ask.
- The protective stop can exit intraday before the channel exit. There is no fixed profit target. After an exit, wait for a later completed-bar entry signal.
Process exits before new entries. Do not close and reverse on the same signal bar in this baseline. These conventions remove ambiguous same-day decisions from the research specification.
Translate the stop into a cash-risk position size
Suppose equity is USD 20,000 and the research budget is 0.25%, or USD 50. A EUR/USD signal has ATR of 0.0040: 40 pips. The initial stop distance is 80 pips. Assuming USD 10 per pip per standard lot and USD 20 per lot for execution-cost allowances, calculated volume is 50 ÷ (800 + 20) = 0.06098 lots.
With a 0.01 lot increment, round down to 0.06 lots, giving an estimated USD 49.20 loss including the stated allowances. Check actual contract specifications and account-currency conversion. Financing is additional for positions held overnight and should be budgeted in the test.
The sample caps portfolio planned stop risk at 1% of equity and one shared directional currency exposure at 0.5%. Scale candidate sizes proportionally when necessary, then round down. If the minimum volume does not fit, skip the signal. These limits are illustrative and cannot contain a gap with certainty.
Follow a hypothetical breakout
Assume the prior 55-day EUR/USD high is 1.1000 and today’s completed close is 1.1020. That is a long signal. If the next session opens with an ask of 1.1024, the strategy enters there, not at the old boundary or yesterday’s close. With ATR of 0.0040, the initial stop is 1.0944.
Suppose the position later remains above its protective stop, but a completed daily close falls below the prior 20-day low. The channel exit is then executed at the next session’s bid. A favourable intraday touch of another level is irrelevant because this baseline does not have a profit target.
If the market gaps below 1.0944 first, the stop may fill lower and the cash loss can exceed the plan. If it merely pierces the 55-day high intraday and closes back inside before entry, this close-confirmed framework creates no trade at all.
Separate the research motivation from proof of this system
A channel breakout is part of the broader trend-following family. The academic time-series-momentum literature provides motivation for studying persistent moves, but it does not certify this 55/20, ATR-stop, retail-cost implementation. Time Series Momentum research.
Keep the entire pipeline reproducible: source prices, bar timezone, signal delay, ATR convention, position rounding, financing and exposure limits. Check several trades by hand, especially the first eligible trade and a bar where an exit boundary changes.
Report losses and no-trade periods as clearly as long trends. A visually attractive chart often selects the one sustained move that makes the rule look obvious. Evaluation must also include the preceding false breaks, the late exits and all currency pairs selected before testing.
Test intrabar uncertainty, costs and parameter stability
Daily candles do not reveal the full order of intraday prices. If a protective stop could trigger within a candle, use suitable lower-timeframe or tick data, or a clearly documented conservative fill assumption. Never choose whichever sequence makes the result most profitable.
Model entry and exit spread, both sides of commission, financing, currency conversion and slippage. Do not grant fills at a channel line after an opening gap. MT5 offers more granular testing modes, but their reliability still depends on data and correct EA logic. MetaQuotes testing modes.
Reserve a final chronological evaluation period. Compare nearby entry lengths such as 50 and 60 as sensitivity checks while retaining the declared baseline. Examine drawdown, turnover, win/loss size, losing streaks and currency concentration. If the finding exists only at one precise parameter or one selected pair, confidence should fall.
Know the failure modes before the first forward test
In a range, price can repeatedly break yesterday’s boundary and then retreat. A short-lived news shock can trigger a signal after most of the move has already occurred. Closing confirmation filters some intraday noise but also delays entries; a slower exit can return a large share of open gains.
Several USD-related pairs can break out together and then reverse together. ATR sizing does not eliminate that concentration, and recent low volatility can increase volume just before conditions change. Add notional and free-margin limits to the cash-stop budget.
Forward-test the exact saved specification, journal rejected and missed trades, and compare realised execution with the model. Pause new entries if data integrity, costs or drawdown depart materially from the test assumptions. The objective is a process whose weaknesses are measurable, not a channel setting that promises to avoid losing trades.
Sources and further reading
Official documentation and research checked for this update. Broker terms depend on the contracting entity and can change.