The thesis
Channel-pullback is a disciplined mean-reversion play on momentum names. It buys pullbacks to the lower regression channel or volume support within confirmed uptrends, then exits at the upper channel or resistance. In plain terms: it tries to buy dips that the broader trend should rescue, and it sells into strength rather than holding for a breakout. The universe is deliberately liquid and large-cap — 24 blue chips spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (KO, PG, COST), healthcare (JNJ, UNH), and energy (XOM, CVX). That is a sensible playground for a channel strategy: these names trend cleanly and rarely gap to zero.
Recent activity
The strategy is live and trading on a scheduled daily cadence. Over the past week it has stayed busy but far from reckless — runs on 2026-09-23 through 09-30 executed between zero and three orders each, with several rejections mixed in (the 09-29 run rejected both of its candidates). Total equity has hovered in a tight band around $10,960–$11,080, closing the 09-30 run at $10,991.89 with $2,055.85 in cash.
The trade tape reflects the thesis well. On 09-28 it sold NVDA at $230.06, then on 09-25 it had bought the same name at $224.16 — a textbook buy-the-dip, sell-the-strength round trip. It rotated out of XOM, MSFT, and BAC into JNJ, KO, MA, and CVX, tilting toward staples and payments. Nothing here looks like style drift.
Performance and validation
This is where the picture turns balanced. The backtest over 451 days shows a 7.62% total return, a 4.19% CAGR, and a 14.83% max drawdown. The Sharpe is a modest 0.40, and the win rate is only 39.39% across 137 trades — meaning the strategy wins less than two times in five but lets its winners outrun its losers. Turnover is a striking 2,311%, so this is an active, fee-sensitive approach; fortunately fees totalled just $137 and FX cost was zero.
Critically, validation did not pass. Three of four walk-forward folds were positive, and the out-of-sample return was a respectable 3.63% at a 0.74 Sharpe. But fold two was ugly: −11.42% with a −1.70 Sharpe and a 16.09% drawdown. The deflated Sharpe ratio (DSR) sits at just 0.196 against a probabilistic Sharpe (PSR) of 0.702, and that gap — stretched across seven trials — is why the gate failed. The headline Sharpe simply isn't robust once you account for how many variants were tried.
The verdict
Strengths: a coherent, trend-aware thesis; clean execution; cheap trading; and genuinely strong folds (fold three returned 20.68% at a 3.86 Sharpe). Risks: a losing-majority win rate, a double-digit drawdown, one deeply negative regime, and a failed validation that says the edge may not generalise. Channel-pullback deserves to keep running live on a small allocation as a data-gathering exercise — but until out-of-sample numbers and the DSR firm up, it is a candidate, not a conviction.