The thesis
donchian-breakout is a classic trend-following system: buy when price clears a 20-day high, exit when it breaks a 20-day low. It trades a universe of 24 US large caps — the usual megacap tech, financials, staples, healthcare and industrials (AAPL, MSFT, NVDA, JPM, V, JNJ, PG, WMT, XOM, CAT, and more). The logic is deliberately simple and asymmetric: accept many small losses, hold the occasional runner. The live metrics bear that signature out — a 38.46% win rate across 108 trades. Most entries are wrong; the strategy relies on the winners being larger than the losers.
Recent activity
The strategy is live and running scheduled sessions. In the last week it executed trades on 2026-10-13 (bought 1 MSFT at $545.66, sold 3 CAT at $767.05) and 2026-10-15 (bought 21 WMT at $118.23, sold 10 NVDA at $217.62). The paper book stood at a total value of $9,769.85 after the 15 Oct run, with $2,241.07 in cash.
The more telling pattern is rejections. Across the six most recent sessions, the strategy flagged 12 rejected orders against only 4 executed — including three straight runs (08–12 Oct) that executed nothing while cash sat at $813.35. Breakout signals were firing, but the book lacked cash to act on them. That is a sizing/liquidity constraint, not a signal problem, and it means live performance may understate what the rules actually called for.
Backtest and validation
Over 451 days the backtest returned 6.95% (CAGR 3.83%, final equity $10,695.11), with a Sharpe of just 0.34 and a max drawdown of 21.73%. Turnover was heavy at 2,082%, though fees stayed low ($108 total). A sub-0.35 Sharpe against a 22% drawdown is a thin risk-adjusted reward.
Walk-forward validation tells a split story. Three of four folds were positive, and the two most recent were genuinely strong — fold 3 returned 14.09% (Sharpe 2.72) and fold 4 returned 11.35% (Sharpe 1.86), both with shallow drawdowns. But fold 2 lost 7.46% (Sharpe −1.07, 15.58% drawdown), and the full-sample Sharpe remains weak.
The validation gate failed. With 6 trials considered, the probabilistic Sharpe ratio sits at 0.674 but the deflated Sharpe ratio is only 0.198 — below the threshold needed to rule out that the result is a product of multiple testing and noise. The strong out-of-sample window is encouraging, but not yet statistically convincing.
Verdict
The strengths are real: a coherent, well-understood trend-following edge, improving recent folds, and disciplined low-cost execution. The risks are equally real: a failed deflated-Sharpe gate, a punishing 22% drawdown, a losing middle fold, and a live book too cash-constrained to take its own signals. This is a strategy worth watching through another out-of-sample window — not one to lean on yet.