The thesis
donchian-breakout is one of the purest ideas in systematic trading: buy a 20-day high, sell a 20-day low. The logic is that markets trend, and breakouts to new highs are the market announcing a trend has begun. It runs across a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (KO, PG, WMT), and healthcare (JNJ, UNH, ABBV) — broad enough that the signal, not stock-picking, is the edge.
Recent activity
The live paper account is quiet but not idle. Over the last week most scheduled runs executed zero trades while rejecting one to three orders — a signature of a breakout system waiting for fresh highs that aren't printing, or bumping into cash constraints. The two active days told the trend-following story cleanly: on 12–13 August it sold laggards (AAPL at 301.65, UNH at 399.25) and bought strength in the payment networks (V at 362.35, MA at 559.85). Earlier rotations followed the same rhythm — into CAT, ABBV, and KO, out of MSFT and CAT.
The scoreboard is sobering: total account value sits around $9,817, below where a fresh paper book would start. The engine is disciplined, but it has not made money on the live tape.
Backtest and validation
Over 451 days the backtest returned +6.95% (CAGR 3.83%, final equity $10,695) with a Sharpe of just 0.34 and a 21.73% max drawdown. The 38.46% win rate is on-brand for trend following — you lose small often and win big occasionally — but the shallow Sharpe against a one-fifth drawdown means the ride is rough for the reward. Turnover north of 2,000% also shows this is an active book, though fees stayed modest at $108.
Walk-forward validation is where it gets interesting — and where it fails the gate. Three of four folds were positive, and the two most recent were genuinely strong: +14.09% (Sharpe 2.72) and +11.35% (Sharpe 1.86). But fold 2 cratered −7.46% (Sharpe −1.07), and the statistics that punish luck-versus-skill are unforgiving: the Deflated Sharpe Ratio is 0.198 and the PSR 0.674, both short of a confident pass across six trials.
The balanced verdict
The strengths are real — a coherent thesis, clean recent out-of-sample behavior, and low costs. The risks are equally real: a full-sample Sharpe that doesn't justify the drawdown, one badly negative regime, and a validation verdict of not passed. The recent folds hint the strategy may be finding its footing, but the numbers say: promising temperament, unproven edge. Watch it; don't size into it.