The thesis
Donchian-breakout is as classic as systematic trading gets: buy when price makes a new 20-day high, exit when it makes a new 20-day low. It is pure trend following, betting that breakouts from large-cap names tend to continue. The universe is 24 blue-chip US stocks — AAPL, MSFT, NVDA, JPM, V, JNJ, XOM, WMT and peers — so there is no small-cap lottery ticket here; the edge, if any, has to come from riding established moves in liquid names.
Backtest profile
Over 451 days the strategy returned 6.95% (final equity $10,695.11 from $10k), a 3.83% CAGR. The risk-adjusted picture is thinner: a Sharpe of 0.34 against a 21.73% max drawdown. The 38.46% win rate across 108 trades looks ugly at a glance, but it is exactly what trend following should produce — most breakouts fail, and the system leans on a minority of winners to carry the book. What gives me more pause is turnover of 2,083%: the strategy churns hard, and while total fees were only $108 with zero FX cost in this run, that churn is a real-world frictional drag that compounds.
Validation: a fail worth reading closely
The walk-forward gate returned passed: false, and the detail is instructive. Three of four folds were positive — fold 3 (+14.09%, Sharpe 2.72) and fold 4 (+11.35%, Sharpe 1.86) were genuinely strong, with the most recent out-of-sample window posting that 11.35% at a healthy 1.86 Sharpe. But fold 2 cratered (−7.46%, Sharpe −1.07) through the early-2025 stretch, dragging the full-period Sharpe down to 0.34.
The statistical flags are the real story. The PSR of 0.674 is lukewarm, and the deflated Sharpe ratio of just 0.198 — which penalises the 6 trials behind this configuration — says we cannot confidently distinguish this edge from luck after accounting for selection. A strong recent half and a weak middle do not add up to a robust strategy; they add up to one that depends heavily on regime.
Live activity
The live book last traded on 29 October, entering ABBV, MA, HD and WMT. Since then it has stalled: every scheduled run from 2–9 November executed 0 trades and rejected 1–2 candidates, with cash pinned at $959.13 and total equity drifting between roughly $10,015 and $10,391. The strategy is fully invested and finding no fresh breakouts clearing its filters — or hitting position limits on the ones it does.
Verdict
Donchian-breakout behaves exactly as its thesis promises, which is both its strength and its warning. The low win rate is by design and the recent folds are encouraging. But a failed validation, a sub-0.2 DSR, heavy turnover and a 21.7% drawdown mean this is a strategy to watch on a short leash, not to size up. I would want to see the recent out-of-sample strength persist through a weaker regime before trusting it with more capital.