The Thesis
Donchian-breakout is trend-following in its most classical form: buy when price sets a new 20-day high, exit when it prints a 20-day low. It trades a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), payments (V, MA), staples (KO, PG, WMT, COST), healthcare, and energy. The logic is deliberately simple and well-documented — it aims to capture sustained moves while cutting losers quickly, accepting many small losses in exchange for a few large winners.
Backtest Performance
Over 451 days the strategy returned 6.95% (final equity $10,695.11), a CAGR of 3.83%, on 108 trades. The numbers read exactly like a trend-follower should: a low 38.46% win rate paired with a positive total return confirms the payoff is asymmetric — the winners more than cover the frequent small losses. The less flattering figures are a 0.34 Sharpe and a 21.73% max drawdown, meaning the return came with meaningful volatility and a deep equity dip. Turnover of 2,083% is high, and while fees totalled only $108 with no FX cost, that churn is a drag worth watching.
Validation: The Red Flag
Validation did not pass, and the detail is nuanced. Across 4 walk-forward folds, 3 were positive, with a healthy aggregate out-of-sample return of 11.35% and OOS Sharpe of 1.86. Folds 3 and 4 were genuinely strong (+14.09% at Sharpe 2.72; +11.35% at Sharpe 1.86). But fold 2 lost 7.46% at a Sharpe of −1.07, and fold 1 was barely positive. The statistics temper the optimism: a probabilistic Sharpe ratio of 0.674 is mediocre, and the deflated Sharpe ratio of 0.198 — which penalises the 6 trials run — says that once we account for selection effects, the edge is not convincingly distinguishable from luck. That is why the gate stayed shut.
Recent Activity
Live behaviour tells its own story. Late September saw real execution — buys in NVDA (10 @ $230.43), CAT, AAPL, WMT, and sells in KO, V, MA, UNH. But every scheduled run from October 2 through October 9 executed zero trades, logging 1–3 rejections each. Cash sits pinned at $813.35 against a portfolio drifting between ~$9,850 and ~$10,093. The pattern strongly suggests the book is fully invested and lacks the cash to act on new breakout signals — the rejections are the strategy wanting to buy and being unable to.
The Balance
Strengths: a coherent, non-overfit premise; genuinely strong recent out-of-sample folds; disciplined, mechanical execution; negligible fee drag.
Risks: a failed validation gate, a deflated Sharpe near zero, a 22% drawdown, and a live book currently too cash-starved to express its own signals. The strategy is doing what trend-followers do — the open question is whether its edge is real or a well-dressed artefact. Until the deflated statistics improve and the cash constraint is resolved, treat it as a promising candidate on probation, not a proven performer.