The thesis
channel-pullback is a textbook mean-reversion-within-trend play. It buys pullbacks to the lower regression channel or volume support in names that are already in confirmed uptrends, then exits at the upper channel or resistance. The universe is 24 large-cap US equities spanning tech, financials, healthcare, staples and energy — AAPL, MSFT, NVDA, JPM, JNJ, XOM and peers. It is live and trading real paper capital on a scheduled daily run.
Recent activity
The strategy is active but not frantic: most sessions execute two orders, with zero rejections across the last six runs. The pattern is clearly rotational — repeated small round-trips in AAPL (buy 5 shares ~$346, sell ~$351, re-buy, sell again) and MSFT (buy 3 ~$538, sell ~$547), plus a fresh PFE position (69 shares at $28.94) and an XOM exit (11 shares at $165.94). That churn is consistent with the backtest's eye-watering 2,311% turnover: this is a high-frequency-of-rotation book, not a buy-and-hold one.
The scoreboard is less flattering. Total account value slipped from $9,989.09 on Nov 5 to $9,870.34 on Nov 12, with cash swinging between roughly $2,570 and $6,384 as positions open and close. A week of small losses is noise, not a verdict — but it is a drift in the wrong direction.
Backtest and validation
Over 451 days the full backtest returned 7.62% (final equity $10,761.52, 4.19% CAGR) on 137 trades. The headline numbers are mixed: a Sharpe of just 0.40, a 14.83% max drawdown, and a 39.39% win rate. A sub-40% win rate with a positive return means the winners are meaningfully larger than the losers — an asymmetric payoff that works, but only if that edge persists.
Here is the problem: validation did not pass. Across four walk-forward folds, three were positive (+6.53%, +20.68%, +3.63%) but fold 2 cratered to −11.42% with a −1.70 Sharpe and a 16% drawdown. The dispersion is enormous — fold 3's 3.86 Sharpe next to fold 2's −1.70 is not a stable signal. Out-of-sample return was 3.63% (Sharpe 0.74), roughly half the in-sample figure. Most damning, the deflated Sharpe ratio sits at 0.196 after adjusting for 7 trials, against a probabilistic Sharpe of 0.702. That gap is the classic fingerprint of a strategy that looks better than it is once you penalise for search.
Verdict
The strength is real: a coherent, well-defined edge that survived three of four regimes and makes money despite losing most individual trades. The risk is equally real: regime fragility, a single catastrophic fold, and a deflated Sharpe that flags overfitting. channel-pullback earns its live slot as an experiment — but it has not cleared the bar for conviction, and the past week's drift is a reminder to watch it closely rather than size it up.