The Donchian channel breakout is the strategy that built the Turtle traders: mark the highest high and lowest low of the last N bars, then buy a close above the upper channel and sell a close below the lower one. It is pure trend-following — you are paying for the privilege of being late, on the bet that the few breakouts that run will pay for all the ones that fail. The classic entry channel is 20 bars, which is the reading we pinned here.
We ran it as a machine would: close beyond the 20-bar high (long) or low (short), a 1.5×ATR(14) stop, a 2R target, on hourly candles built from 12 months of real Binance 1-minute data on BTC and ETH — around the clock, both directions, 1% risk, 0.05% commission per side, the same deterministic engine that powers Secuora’s AI backtester. The honest headline: the raw rule lost on both symbols, but the two results were not close to each other. ETH finished close to breakeven on profit factor (just under 1.0) (profit factor just under 1.0) while BTC bled meaningfully more — a clean illustration that a breakout system is only as good as the trends the instrument happens to print. Full numbers below.
Verified Result
Weak: the mechanical version barely cleared, or failed to clear, the fee hurdle.
| Market | TF | Trades | Win | PF | Max DD | Net |
|---|---|---|---|---|---|---|
| ETH | 1h | 389 | 35.5% | 0.97 | 27.3% | -9.7% |
| BTC | 1h | 414 | 32.4% | 0.76 | 52.2% | -51.1% |
How the SVS 39 breaks down ▾
12 months of real 1-minute data, fees on (0.05%/side), $10k start, 1% risk. How the score works →
The exact rules we tested
- Compute the 20-bar Donchian channel on hourly candles: the highest high and lowest low of the prior 20 bars.
- Go long on a candle that closes above the upper channel; go short on a candle that closes below the lower channel.
- Cross semantics — the breakout fires once as price leaves the channel, not on every bar that stays beyond it.
- Stop 1.5×ATR(14) from entry; target 2R.
- No session filter — crypto trades 24/7, so every qualifying channel break is taken, both directions.
- Risk 1% of equity per trade; 0.05% commission per side; 10× max notional leverage.
Results
Binance spot 1-minute klines (data-api.binance.vision), aggregated per strategy timeframe · starting balance $10,000 · risk 1%/trade · Commission 0.05% per side; no spread/slippage modeled (BTC/ETH spot spreads are sub-basis-point); position size capped at 10× notional leverage. Generated 2026-06-12 by the Secuora ai-strategy deterministic runner (same engine as the in-app AI backtester).
| Month | Trades | Win rate | Net P&L |
|---|---|---|---|
| 2025-06 | 36 | 31% | −$828 |
| 2025-07 | 36 | 22% | −$1,517 |
| 2025-08 | 30 | 33% | −$356 |
| 2025-09 | 34 | 29% | −$751 |
| 2025-10 | 42 | 40% | $224 |
| 2025-11 | 35 | 34% | −$172 |
| 2025-12 | 36 | 25% | −$847 |
| 2026-01 | 41 | 41% | $162 |
| 2026-02 | 33 | 36% | −$7 |
| 2026-03 | 31 | 39% | $120 |
| 2026-04 | 29 | 17% | −$968 |
| 2026-05 | 31 | 35% | −$171 |
| Month | Trades | Win rate | Net P&L |
|---|---|---|---|
| 2025-06 | 29 | 41% | $455 |
| 2025-07 | 37 | 35% | −$122 |
| 2025-08 | 37 | 43% | $877 |
| 2025-09 | 27 | 33% | −$323 |
| 2025-10 | 40 | 40% | $520 |
| 2025-11 | 34 | 26% | −$1,020 |
| 2025-12 | 27 | 30% | −$535 |
| 2026-01 | 36 | 53% | $1,805 |
| 2026-02 | 34 | 35% | −$55 |
| 2026-03 | 27 | 37% | $104 |
| 2026-04 | 28 | 25% | −$1,084 |
| 2026-05 | 33 | 21% | −$1,588 |
Assumptions (how loose terms were pinned down)
- Close beyond the 20-bar high (long) or low (short); 1.5x ATR stop; 2R target
Same rule, two instruments, very different outcomes
Run the identical 20-bar breakout on BTC and ETH for the same 12 months and the gap between them is the whole lesson. ETH landed close enough to breakeven that a slightly different fee tier or a single avoided whipsaw would flip the sign; BTC lost several times as much and spent more than half its peak in drawdown along the way. Nothing about the rule changed — only the trend texture of the instrument did. That is the structural truth of every breakout system: it does not have a win rate so much as it has a dependence on whether the market trended while you held it. A sub-35% win rate is normal and fine when the 2R winners are large enough; the months that hurt are the balance phases, where every channel break reverses back through the range and the 1.5×ATR stop pays for it.
The honest read is that this is the floor, not the verdict. There is no trend filter, no volatility regime gate, no attempt to skip the chop — one rule, every break taken. The original Turtle system added a longer exit channel, position pyramiding and a portfolio of uncorrelated markets precisely because the bare entry, on one instrument, is a coin flip with costs attached. The monthly table shows which months carried the year and which gave it back, and that texture is the thing to study before you add a single filter.
How to backtest the Donchian breakout on Secuora
A channel breakout is a built-in primitive of the AI backtester, so this strategy automates end to end — describe it once and re-run it across instruments to see the dependence for yourself.
- Open /backtest/ai and describe the rule in plain English: "go long on an hourly close above the 20-bar high, short on a close below the 20-bar low, 1.5×ATR stop, 2R target." It compiles to the same deterministic engine that produced this page.
- Run it on BTC and ETH with costs on and confirm you reproduce the split above — one near breakeven, one clearly losing. Reproducing the shape is what separates research from a screenshot.
- Change one variable per run: widen the channel to 55 bars (the Turtle long-term entry), add a higher-timeframe trend filter so you only break in the trend’s direction, or swap the 2R target for a trailing channel exit.
- Sign up free and use the replay terminal — open the chart, draw the 20-bar channel, and replay a ranging month bar by bar with simulated stop and limit orders to feel how many breaks failed before one ran.
- Journal the variants worth trading — rules followed, confluences, screenshots — so the version you risk money on is the version you actually tested.
Methodology, in one paragraph
Data: Binance spot 1-minute klines (data-api.binance.vision), aggregated per strategy timeframe, June 1, 2025 – June 1, 2026 (12 months). Execution: Secuora’s deterministic strategy runner (the same engine behind the in-app AI backtester) — single position at a time, entries at the close of the signal candle, commission 0.05% per side; no spread/slippage modeled (btc/eth spot spreads are sub-basis-point); position size capped at 10× notional leverage, starting balance $10,000, 1% risk per trade. Swings are confirmed fractals with no look-ahead. These are mechanical results: no discretion, every signal taken. Past performance does not predict future results; this is research, not financial advice.
Frequently asked questions
What is the Donchian channel breakout strategy?
Mark the highest high and lowest low of the last N bars (commonly 20) to form a channel, then trade the break: buy a close above the upper band, sell a close below the lower band. It is a pure trend-following entry — the same idea that powered the Turtle traders — and it relies on the occasional large trend paying for a long string of small failed breaks.
What win rate does the Donchian breakout have?
In our 12-month hourly test the win rate sat in the low-to-mid 30s on both BTC and ETH — and that is normal for a 2R breakout, which only needs winners roughly twice the size of losers to break even. Judge the expectancy and the drawdown, not the win rate alone: ETH finished near breakeven while BTC lost meaningfully more, on the identical rule, because the trends differed.
Does the Donchian breakout work on crypto?
Not in this raw, unfiltered form. The bare 20-bar break lost on both BTC and ETH over 12 months after costs, though ETH came close to breakeven on profit factor (just under 1.0). Breakout systems live or die on trend persistence, so the realistic path is fewer, better-located breaks — a longer channel, a trend filter, a regime gate — which is exactly what this unfiltered baseline lets you measure against.
How do I backtest the Donchian breakout myself?
Two ways on Secuora: describe it in plain English at /backtest/ai — channel breakout is a built-in primitive and it compiles to the same engine that produced this research — or sign up free and use the replay terminal to draw the 20-bar channel and trade the breaks bar by bar with simulated stop and limit orders.