The thesis
channel-pullback is a disciplined dip-buyer. It waits for a confirmed uptrend, then buys pullbacks to the lower regression channel or a volume-support level, aiming to exit at the upper channel or overhead resistance. The universe is a conservative 24-name roster of megacap tech, financials, staples, healthcare, and energy — AAPL, MSFT, NVDA, JPM, V, JNJ, XOM and peers. It is a sensible framing: buy strength on weakness, sell into strength. The strategy is currently live.
Recent activity
Activity is steady rather than frantic. Over the last two weeks the strategy logged scheduled runs on most trading days, executing between zero and three orders each time. Recent fills include buying BAC (32 shares at $59.86) while trimming NVDA (9 shares at $212.03) on 2026-09-15, and rotating through HD, V, AAPL and UNH earlier in the week. The BAC round-trip is instructive: it sold 30 shares at $62.79 on 09-09, then rebought 32 at $59.86 six days later — exactly the buy-the-pullback behaviour the thesis promises.
That said, the live book has drifted lower. Total portfolio value slipped from $9,466.89 on 09-08 to $9,341.25 on 09-15 — a small but real drawdown over a window where cash swung from near-zero to nearly $5,700 and back. Execution is clean (very few rejections), but the recent tape has not rewarded the entries.
Backtest and validation
Over 451 days the backtest returned 7.62% (final equity $10,761.52, CAGR 4.19%) on a Sharpe of just 0.40, with a 14.83% max drawdown and a 39.39% win rate across 137 trades. A sub-40% hit rate that still nets positive means the winners are meaningfully larger than the losers — respectable expectancy, but psychologically demanding. Turnover is a heavy 2,311%, so the strategy is churning; fees ($137) stay modest only because the paper book is small.
The validation verdict is the headline: it fails. Walk-forward across four folds shows three positive, but the dispersion is alarming — Fold 3 returned +20.68% (Sharpe 3.86) while Fold 2 lost 11.42% (Sharpe -1.70, 16% drawdown). Out-of-sample return averages a thin 3.63% (OOS Sharpe 0.74). Critically, with 7 trials the deflated Sharpe ratio is 0.196 — far below the confidence threshold — even though the probabilistic Sharpe (0.702) looks superficially fine. The DSR is telling us the edge may be an artifact of selection, not a durable signal.
Verdict
channel-pullback is a coherent, well-behaved idea that executes its thesis faithfully. But the numbers counsel caution: a modest full-period return, a middling Sharpe, and — most importantly — a formal validation failure driven by fold-to-fold instability and a weak deflated Sharpe. Fold 2 shows how quickly a mean-reversion entry becomes a falling-knife catch when the trend breaks. Until out-of-sample consistency improves, this belongs on the watch-and-refine list, not the conviction shelf.