The thesis
donchian-breakout is a textbook trend-follower: buy 20-day high breakouts, exit on 20-day lows. It runs across a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), healthcare (UNH, JNJ, PFE), staples (PG, KO, WMT), and industrials/energy (CAT, XOM, CVX). The bet is simple and well-worn — that price momentum persists long enough to pay for the many small losses trend systems accumulate while waiting for a runner.
Backtest performance
Over 451 trading days the strategy returned 6.95% (CAGR 3.83%), ending at $10,695 of equity on $10,000. That headline hides a rough ride: a 21.73% max drawdown, a Sharpe of just 0.34, and a 38.46% win rate across 108 trades. The low hit rate is expected for breakout logic — profitability depends on asymmetric payoffs, not on being right often. More concerning is turnover of 2,083%: the book churns aggressively, and while total fees ($108) are modest here, that pace erodes edge in any friction-heavier environment.
Validation: three of four, but no pass
The walk-forward study is where the picture gets interesting. Across four folds the strategy was positive in three, and the two most recent windows are its best: fold 3 returned +14.09% (Sharpe 2.72) and fold 4 +11.35% (Sharpe 1.86). The out-of-sample composite Sharpe of 1.86 is genuinely strong. But fold 2 (Jan–Jul 2025) lost 7.46% with a negative Sharpe, the kind of whipsaw-heavy regime that punishes breakout systems.
Despite the recent strength, the gate reads validation: failed. The deflated Sharpe ratio sits at just 0.198 and PSR at 0.674 — after accounting for six trials, the evidence isn't strong enough to rule out luck. That is the correct, conservative call: recent form is encouraging, but not yet statistically clean.
Recent activity: stalled
The live book tells a cautionary story. The last executed trades were in early-to-mid June (buys in KO, UNH, ABBV, AAPL; sells in CAT, MSFT). Since late July, every scheduled run has printed the same line: 0 executed, 2 rejected, with cash frozen at $1,760.89. Meanwhile total equity has drifted down from ~$10,406 (Jul 29) to $9,849 (Aug 3) — the erosion is coming from existing holdings, not new positioning. The strategy is effectively parked, unable to place its intended orders while its book bleeds.
Verdict
There's a credible signal here — the out-of-sample folds are real strengths, and the logic is sound. But the risks are equally real: a sub-0.35 full-period Sharpe, a 22% drawdown, heavy turnover, and a validation gate that (rightly) refused to pass on six trials of data. Layer on the current operational stall, and the honest read is a promising but unproven system. It deserves continued paper-trading and a look at why orders keep getting rejected — not yet real capital.