The thesis
The donchian-breakout strategy runs one of the oldest ideas in systematic trading: buy 20-day high breakouts and exit on 20-day lows. It's pure trend following, applied to a 24-name universe of large-cap US equities spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (KO, PG, WMT), and healthcare (JNJ, UNH). The premise is that a stock printing a new one-month high is more likely to keep rising than reverse — and that cutting losers at new lows keeps drawdowns bounded.
Backtest performance
Over a 451-day backtest, the strategy returned 6.95% (a 3.83% CAGR), ending at $10,695 on a $10,000 base. The character is textbook trend following: a 38.46% win rate across 108 trades, meaning most positions lose small while a minority of winners carry the book. That asymmetry is a feature, not a bug — but it demands discipline, because the max drawdown reached 21.73% against a Sharpe of just 0.34. Turnover was extreme at 2,082%, so the $108 in fees is a real, if modest, drag.
What validation says
Here's where the picture gets honest. Our four-fold walk-forward test did not pass. Three of four folds were positive, and the two most recent were genuinely strong — fold 3 returned 14.09% at a 2.72 Sharpe, fold 4 returned 11.35% at a 1.86 Sharpe. But fold 2 (Jan–Jul 2025) lost 7.46% at a -1.07 Sharpe, exactly the sideways, whipsaw regime that punishes breakout systems.
The deeper red flag is the statistics. The Probabilistic Sharpe Ratio sits at 0.674 and the Deflated Sharpe Ratio at just 0.198 — the DSR adjusts for the 6 trials run, and a value that low means we can't confidently reject the possibility that the headline Sharpe is a product of selection rather than skill. A strong recent out-of-sample stretch is encouraging, but not enough to clear the gate.
Recent activity
The live account is quiet — arguably too quiet. The last six scheduled runs (July 22–29) all executed zero trades, each rejecting one or two candidate orders. Equity has drifted around $10,100–$10,400 with $1,760.89 in idle cash. The last real fills were back in early June: buys in KO, UNH, ABBV, AAPL, and CAT, plus trims in MSFT and CAT. A breakout system going dormant usually means the universe simply isn't printing fresh 20-day highs — consistent with a market that's chopping rather than trending.
The verdict
donchian-breakout is doing what it's supposed to: participating in trends, cutting losers, and sitting on its hands when there's nothing to buy. The recent out-of-sample folds are legitimately promising. But the failed validation, the sub-0.35 Sharpe over the full sample, a 21%+ drawdown, and the low deflated Sharpe all argue for keeping this one on a short leash. It's a candidate to watch, not yet a strategy to trust with size.