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 (JNJ, UNH), and staples (PG, KO, WMT). The logic is deliberately simple — let winners run, cut losers when price rolls over. The strategy is currently live.
Recent activity
The last week has been quiet, and not in a good way. Six consecutive scheduled runs from 17–24 August each executed zero trades and rejected one or two candidate orders. Cash has been frozen at $1,510.88 the entire stretch while total equity drifted from $9,755 up to $10,229 — meaning the book's existing positions, not new signals, are driving the P&L. The repeated rejections suggest sizing or cash constraints are blocking fresh breakouts from being filled; worth investigating before more signals are missed.
The last genuine activity was mid-August: buys in V (5 @ $362.35) and MA (3 @ $559.85), paired with exits in UNH and AAPL. Earlier executed trades in June (KO, CAT, ABBV, UNH) show the rotation working as designed when cash and signals align.
Backtest and validation
Here is where the picture gets nuanced. The full backtest over 451 days returns +6.95% (CAGR 3.83%), with a Sharpe of just 0.34 and a stiff 21.73% max drawdown. Win rate is 38.46% across 108 trades — low, but expected for trend following, where a minority of large winners is supposed to pay for many small losses. Turnover is a heavy 2,083%.
The walk-forward folds tell a more optimistic story at the edges. Of four out-of-sample folds, three were positive. Fold 3 (+14.09%, Sharpe 2.72) and the most recent Fold 4 (+11.35%, Sharpe 1.86, shallow 6% drawdown) are genuinely strong. The blemish is Fold 2 (−7.46%, Sharpe −1.07) during early-to-mid 2025 — a choppy, mean-reverting regime that punishes breakout systems.
Why validation still failed
Despite those recent folds, the formal gate reads passed: false. The reason is in the multiple-testing math: with 6 trials, the Probabilistic Sharpe Ratio sits at 0.674, but the Deflated Sharpe Ratio is only 0.198 — well below any confidence threshold. Once you haircut the full-sample Sharpe of 0.34 for the number of configurations tried, the edge is not statistically distinguishable from luck.
The balanced read
The strengths are real: a clean, well-understood thesis, an improving recent trajectory, and controlled drawdowns in the latest folds. But the risks are equally real — a sub-0.5 Sharpe overall, a punishing Fold 2, and a validation failure that says the recent gains may not survive out-of-sample scrutiny. Add the week of rejected orders, and the honest verdict is: promising regime-dependent behaviour, not yet a proven edge. It stays live on a short leash, and the order-rejection issue deserves a look first.