The thesis
Donchian Breakout is a classic trend-following rule: buy when price sets a fresh 20-day high, exit when it prints a 20-day low. It runs across a 24-name universe of US large caps spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (KO, PG, WMT), health care (UNH, PFE, ABBV) and industrials (CAT, HON). The logic is deliberately simple and mechanical — it makes no forecast, only follows confirmed momentum and cuts positions that roll over.
Backtest performance
Over 451 days the strategy returned 6.95% (final equity $10,695.11 on a $10k base), a 3.83% CAGR, with a Sharpe of 0.34 and a steep max drawdown of 21.73%. Win rate was 38.46% across 108 trades — low, but not alarming for trend-following, where a minority of large winners is expected to carry a majority of small losers. Two figures deserve scrutiny: turnover of 2,082% is very high and implies the edge must survive real trading friction, and total fees of $108 across 108 trades confirm the churn. The Sharpe-to-drawdown profile is the honest weakness here — you are paying with a fifth of capital in drawdown for a mid-single-digit return.
Validation: the gate said no
Walk-forward validation is more encouraging than the headline backtest, yet the gate still failed. Across four folds, three were positive: fold 3 returned 14.09% (Sharpe 2.72) and fold 4 11.35% (Sharpe 1.86, and the reported out-of-sample slice), both with shallow drawdowns near 3–6%. The blemish is fold 2 (Jan–Jul 2025): −7.46% at a Sharpe of −1.07 with a 15.58% drawdown — a whipsaw regime that breakout systems historically hate. The deflated statistics tell the real story: PSR 0.674 is respectable, but DSR 0.198 is weak once you account for 6 trials of selection. In plain terms, some of the apparent edge is likely luck from testing multiple variants. That is why the gate withheld its blessing despite three green folds.
Recent activity: live, but idling
The strategy is live and trading paper capital, but the last week has been quiet. Scheduled runs from Aug 28 through Sep 4 each executed zero orders while rejecting one or two candidates per session — the breakout filter simply is not triggering, or position/cash limits are blocking entries. The paper book sits around $9,975–$10,084 total with $1,510.88 in cash, essentially flat. The most recent fills were back on Aug 12–13 (buys in V and MA, sells in AAPL and UNH), with a prior cluster in early June.
Verdict
Donchian Breakout is a coherent, transparent trend follower with genuinely strong good-regime folds — but a failed validation gate, a 21.73% drawdown, and a fold-2 collapse are real caution flags. The current run of all-rejected sessions suggests it is patiently waiting for trends rather than forcing trades, which is faithful to the thesis. Watch whether the OOS strength persists or whether the deflated Sharpe was warning us correctly.