The thesis
donchian-breakout is a pure trend-following system: buy when a name in its 24-stock large-cap universe (AAPL, MSFT, NVDA, JPM, V, XOM and peers) prints a fresh 20-day high, and exit when it prints a 20-day low. It is one of the oldest ideas in systematic trading, and it comes with a well-known personality — many small losses punctuated by a few large winners. The reported 38.46% win rate is not a bug; it is exactly what a breakout system should look like. The question is always whether the winners are big enough to pay for the losers and the churn.
Backtest performance
Over 451 days and 108 trades, the backtest returned 6.95% (a 3.83% CAGR), ending at $10,695 on a $10,000 base. The risk-adjusted picture is where enthusiasm should cool: a Sharpe of 0.34 is thin, and the 21.73% max drawdown is a real gut-check for that level of return. Turnover of 2,082% is enormous — the book effectively recycles itself twenty times over — yet total fees were only $108 and FX cost was zero, so trading friction is not the primary drag here. The core issue is that the strategy is taking meaningful drawdown risk for a modest, single-digit reward.
Validation: it did not pass
The walk-forward validation failed, and the nuance matters. Three of four folds were positive, and the out-of-sample legs actually look strong on the surface — fold 3 returned +14.09% (Sharpe 2.72) and fold 4 +11.35% (Sharpe 1.86). But fold 2 lost -7.46% with a -1.07 Sharpe, showing how brutal a trendless, choppy regime is for breakouts. More importantly, the deflated Sharpe ratio (DSR) is just 0.198 against 6 trials: once you penalise for the number of variants tested, the edge is not statistically convincing. PSR of 0.674 is only mild support. This is validation working as intended — flagging that the encouraging out-of-sample folds may owe as much to a strong trending market as to genuine skill.
Recent activity: idling
The live tape is the most concrete concern. The last executed trades were on 2026-08-13 (buy V, sell UNH). Since then, every scheduled run from 2026-09-01 through 2026-09-08 reads the same: 0 executed, 1–2 rejected, cash pinned at $1,510.88. No new breakouts are clearing whatever entry gate is in place, so the strategy is passively holding. Meanwhile total equity has slid from $10,084 (Sep 3) to $9,837 (Sep 8) — a mark-to-market bleed on open positions, not a trading loss. A breakout system that stops finding breakouts and just sits in drifting holdings is not doing its job.
Verdict
The design is sound and the out-of-sample folds are genuinely interesting, but the honest read is not yet proven: a failed validation, a weak deflated Sharpe, and a live book that is currently stalled with rejected orders. This is a strategy to keep on a short leash and watch, not to size up.