Unlock This Note

ETH gained more than 50% in three years and lost two thirds of its value along the way. This minute-level breakout strategy only rides the upward bursts and sits in cash the rest of the time. The Donchian channel is one of the oldest trend-following tools: track the highest high and lowest low over a lookback window, and chase price when it breaks above the top. This ETH strategy (ETH Donchian Breakout, listed by a community seller) runs it on minute bars: the channel is built from the last 175 one-hour bars, a 1-minute close above the upper band opens a long on ETH perpetuals, and the trade exits when price falls back to the channel midpoint or hits a 1.25% stop or a 4% trailing stop. It never shorts. Backtested September 2023 to October 2026: +90.1% total return, about 22.9% annualized, Sharpe 1.68, max drawdown −10.7%. Holding ETH over the same period returned +56.9% with a −67.6% drawdown. How does a Donchian breakout work on ETH minute bars? The channel uses hourly bars: the highest high of the previous 175 completed hours is the upper band, the lowest low is the lower band, and their midpoint is the re-arm line. Entries are checked on 1-minute closes: a close above the upper band opens a long only if the 14-period ADX is at least 25 and the bar's volume is no lower than the past hour's average. Four exits: a close back below the midpoint, a 1.25% stop loss, a 4% trailing stop from the high-water mark, and a stop moved to breakeven once the trade is up 0.5%. After an exit it does not chase: price has to return to the midpoint and break the upper band again before a new signal counts. That makes it a quiet strategy. It entered 76 times in three years, about 25 a year; half the trades were over within two and a half hours, and it was in the market on roughly one trading day in ten. Why is the drawdown a sixth of buy-and-hold's? Because it is out of the market most of the time. ETH topped $4,000 in March 2024 and slid to $1,400 by April 2025, a two-thirds drop for anyone holding. The strategy's worst drawdown over the same stretch was −10.7%, and its worst year was still positive. The price is that it also catches only part of the rallies: ETH rose about 40% in each of 2023 and 2024, and the strategy made 14.4% and 7.7%. Backtest performance (Sep 2023 to Oct 2026, 1-minute bars) MetricValueTotal return+90.1% (buy-and-hold +56.9%)Annualized~22.9% (buy-and-hold ~15.6%)Sharpe1.68Max drawdown−10.7% (buy-and-hold −67.6%)Trades76 round trips, 35.5% win rate, average winner +3.65%, average loser −0.56%By year (strategy vs buy-and-hold)2023 +14.4% vs +40.3%, 2024 +7.7% vs +41.7%, 2025 +34.4% vs −11.6%, 2026 to date +14.8% vs −14.9%MCPT p-value< 0.001 (none of 1,000 random shuffles did better) Includes a 0.05% fee per side (Binance USDT-M perpetual taker rate); fees consumed 7.6% over three years. Stops and trailing stops trigger on the bar's high or low and fill at the next bar's open. Funding payments are not included. Backtest results, not a guarantee of future performance. Three things to know before you run it First, the profit is concentrated in a handful of trades. The five best of the 76 trades add up to more than the entire gross profit, and the single trade on May 8, 2025 made 27.6%. In 2024 the strategy won only 4 of 24 trades and the whole year rested on one trade in February. The equity curve looks like a staircase, flat most of the time, so you need to be able to wait. Second, the 175-bar channel length was picked on the same data. Move it to 150 or 200 and Sharpe drops to 1.44 and 1.63; at 125 it is 0.72. The stop, trailing stop and breakeven settings are stable, staying between 1.5 and 1.8 wherever you move them. It passes our robustness gate (neighborhood average ÷ chosen cell = 0.90), but there is no out-of-sample period, so expect live results below the backtest. Third, the ADX and volume filters barely matter in this sample: however you set the thresholds, the same trades come through. They are a fuse, not the source of returns. The rule is simple. Watching it every minute is the work. The logic is all above. What takes time is recomputing the channel and filters every minute, tracking the stop and trailing stop, and placing orders without missing a bar. Blave Agent runs that chain for you: get this strategy from the library, then tell your Agent to deploy it. See the ETH Donchian breakout strategy in the library →