Donchian Breakdown Short

Donchian Breakdown Short trades one idea: in a market that is already falling, a break of the N-bar low tends to keep going. How it works. The entry channel is the lowest low of the last 16 candles on the 1h timeframe, and a close beneath that level is the breakdown signal. Two regime filters decide whether the break is worth taking: price must be trading below a 66-period trend EMA, and that EMA must itself be lower than it was 4 candles earlier. Together they keep the strategy out of the sharp counter-trend bounces that produce most false breaks. The position is covered on the opposing side of a deliberately slower 42-candle channel, which gives a genuine breakdown room to run instead of being shaken out by the first rally. Market and risk. Short-only, futures with isolated margin, 1h candles, six USDT-margined perpetuals (BTC, ETH, ATOM, DOT, LINK, XRP). A hard -10% stoploss is always in force, because a short position carries unbounded loss if price rallies against it. Maximum three concurrent positions at 10 USDT of stake each. Results. Over the 285-day in-sample window (2025-08-29 to 2026-06-09) it returned +2.34% while the market itself fell 61.92%, from 184 trades at a 49.5% win rate, profit factor 1.61, Sharpe 2.44 and a maximum drawdown of 0.67%. On the 71-day out-of-sample leg that followed, which it was never tuned against, it returned +0.28% with profit factor 1.44 and Sharpe 1.55 across 40 trades. Tuning. The four signal parameters (entry channel, exit channel, trend EMA period, trend slope lookback) are exposed as hyperopt parameters and were fitted by a 100-epoch Sharpe-ratio search over the in-sample leg only, restricted to the buy and sell spaces. ROI and stoploss were deliberately excluded from that search, so the risk settings are the author's choice rather than a fitted artefact. What to expect. This is a directional short strategy built for a falling market. It made money through a severe downtrend; it is not designed for a rising one and will find few qualifying signals there. Absolute returns are modest because position sizing is conservative relative to the wallet - the edge shows in the profit factor and in the gap to buy-and-hold, not in the headline percentage.