The thesis
Donchian-breakout is a purist trend-following system: buy a 20-day high, exit on a 20-day low, and let the channel do the timing. It runs across a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, WMT) and healthcare (JNJ, UNH). The logic is deliberately dumb in the best sense — no forecasting, just riding sustained moves and cutting the rest.
Recent activity
The live book is quiet, and that is the headline. Across the six most recent scheduled runs (20–27 August), the strategy executed zero trades and rejected one or two candidate orders each day. Cash has sat frozen at $1,510.88, with total paper equity drifting between roughly $10,070 and $10,260 — small mark-to-market wiggle on existing positions rather than fresh conviction. The last real fills were back on 12–13 August: buying V and MA while trimming AAPL and UNH.
A week of rejected signals isn't necessarily broken behaviour — it can mean breakout candidates appeared but failed a sizing or risk check. Still, a trend-follower that stops adding when it can't get filled is a system to watch, not celebrate.
Backtest and validation
The backtest tells an honest, unglamorous story. Over 451 days and 108 trades it returned 6.95% (3.83% CAGR) with a Sharpe of just 0.34 and a maximum drawdown of 21.73%. Win rate is 38.46% — entirely expected for trend following, where a minority of trades carry the whole return. Turnover of 2,083% is high, though at $108 total fees the cost drag is modest.
The walk-forward validation is the more interesting document, and it did not pass. Three of four folds were positive, and the out-of-sample stretch looks genuinely good — 11.35% return at a 1.86 Sharpe in the most recent window, and a stellar fold 3 (14.09%, Sharpe 2.72). But fold 2 was ugly: -7.46% at a -1.07 Sharpe through early 2025, a choppy, trendless regime that punishes breakout systems. Crucially, after adjusting for the six trials run, the deflated Sharpe ratio is 0.198 — the probabilistic Sharpe of 0.674 doesn't survive the multiple-testing haircut.
Verdict
Strengths: the edge is real in trending regimes, the drawdowns per fold are contained, and the OOS behaviour is encouraging. Risks: whipsaw regimes gut it (fold 2), the aggregate Sharpe is thin, and the deflated statistics say we haven't cleared the overfitting bar. The recent run of rejected orders is a second yellow flag. Live-but-unproven is the fair label — keep it small, and let the next few trending weeks make the case the validation gate wouldn't.