The thesis
The donchian-breakout strategy is about as orthodox as trend-following gets: buy a 20-day high breakout, exit on a 20-day low. It runs long-only across a 24-name universe of US large caps spanning tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, ABBV), and staples/industrials (PG, KO, WMT, CAT, XOM). The idea is old and well-studied — ride confirmed momentum, cut positions when the channel breaks down. The strategy is currently marked live.
Recent activity
Most of the recent tape is notably quiet. The Season 2 reset on 29 October was the one active day: four buys executed — ABBV (8 @ $289.91), MA (3 @ $636.60), HD (7 @ $328.57), and WMT (20 @ $121.17) — deploying the opening capital. Since then, every scheduled run has placed zero trades, with one or two orders rejected each session (2 November through 5 November each logged "0 executed, 2 rejected"). Cash has been pinned at $959.13 throughout, which strongly suggests the rejections are capital- or sizing-constrained rather than signal-driven: the breakout logic may be firing, but there's nothing left to fund the orders. Total account value has drifted between roughly $9,912 and $10,391, sitting at $10,296.64 on 5 November — modestly above the $10k reset line.
Backtest and validation
The backtest covers 451 days and 108 trades, ending at $10,695.11 for a 6.95% total return (3.83% CAGR). The profile is textbook trend-following: a low 38.46% win rate carried by a smaller number of larger winners. The weaknesses are equally textbook — a Sharpe of just 0.34, a 21.73% max drawdown, and eye-watering 2,082% turnover, though fees stayed trivial ($108 total, no FX cost).
Validation did not pass, and the fold data explains the tension. Three of four walk-forward folds were positive, with the two most recent windows strong: fold 3 returned +14.09% (Sharpe 2.72) and fold 4 — the out-of-sample window — returned +11.35% (Sharpe 1.86) on a contained 6.01% drawdown. But fold 2 (Jan–Jul 2025) lost 7.46% with a −1.07 Sharpe, the classic whipsaw tax when trends fail to form. The deciding numbers are the statistical ones: PSR of 0.674 is borderline, but the Deflated Sharpe Ratio of 0.198 — adjusted for the 6 trials run — says the full-sample edge is not convincingly distinct from luck once we account for the search.
The verdict
The strategy's recent out-of-sample behaviour is genuinely encouraging, and the construction is clean and interpretable. But a 0.34 full-sample Sharpe and a sub-0.2 DSR are exactly what the validation gate exists to catch. Combine that with a 22% drawdown and live runs that currently can't fund their signals, and this reads as a promising but unproven breakout engine — worth watching in the favourable regime it's enjoying now, not worth trusting the capital to yet.