The thesis
Donchian-breakout is trend-following in its most classic form: buy a 20-day high, sell when price prints a 20-day low. It trades a 24-name universe of large-cap US equities — the mega-cap tech and financials (AAPL, MSFT, NVDA, JPM, V, MA) alongside defensive staples and healthcare (JNJ, UNH, KO, PG, WMT). The premise is old and honest: let winners run, cut losers at the channel, and accept a low hit rate in exchange for the occasional trend that pays for everything else.
Recent activity
The last executed fills came in mid-August — a rotation that bought V (5 @ $362.35) and MA (3 @ $559.85) while selling AAPL (7 @ $301.65) and UNH (3 @ $399.25). Since then the strategy has gone quiet. Every scheduled run from September 10 through 17 reports 0 executed, 1–2 rejected, with cash pinned at $1,510.88 and total equity oscillating narrowly between roughly $9,851 and $10,025. In other words, the model is generating signals but the orders aren't clearing — most likely position-sizing or cash constraints against a nearly fully invested book. The portfolio is effectively flat, hovering around its $10k starting line.
Backtest and validation
Over 451 days the backtest returned 6.95% (final equity $10,695), a 3.83% CAGR, on a 0.34 Sharpe and a 21.73% max drawdown. The win rate is 38.46% across 108 trades — low, but entirely expected for breakout systems. Turnover is the number that stands out: 2,082%, a heavy churn that racked up $108 in fees.
The walk-forward picture is more encouraging than the headline. Three of four folds were positive, and the two most recent were genuinely strong — fold 3 (+14.09%, Sharpe 2.72) and fold 4 (+11.35%, Sharpe 1.86). The blemish is fold 2 (Jan–Jul 2025): -7.46% at a -1.07 Sharpe, a choppy stretch where breakouts kept failing and the channel whipsawed the book.
Despite the recent out-of-sample Sharpe of 1.86, validation failed. The Probabilistic Sharpe Ratio sits at 0.674, but the Deflated Sharpe Ratio — which penalises for the 6 trials behind this configuration — collapses to 0.198, well short of the confidence we require to certify an edge. The full-sample 0.34 Sharpe simply isn't robust enough once multiple-testing is accounted for.
The balance
Strengths: a transparent, well-understood rule set; strong recent OOS performance; and no reliance on curve-fit parameters beyond the canonical 20-day window.
Risks: a 21.73% drawdown and sub-0.4 aggregate Sharpe make this a bumpy ride for thin returns; the fold-2 breakdown shows real vulnerability to sideways regimes; and the current run of rejected orders suggests the live implementation isn't executing its own signals. Until the deflated Sharpe clears the gate, donchian-breakout stays a monitored candidate — promising in the right regime, unproven across all of them.