The thesis
Channel-pullback is a mean-reversion idea wearing a trend-following jacket. It buys pullbacks to the lower regression channel or volume support — but only inside confirmed uptrends — and exits at the upper channel or resistance. In principle this filters out the classic value-trap of the falling knife: you're only fading dips within markets already going the right way. It runs across a 24-name large-cap universe spanning tech, financials, healthcare, staples, and energy, which gives it room to diversify entries.
Recent activity
Trading has been light and orderly. Over the last week, scheduled runs executed between zero and two orders each, with no rejections — a sign the execution plumbing is healthy even when conviction is low. Two round-trips stand out. A BAC buy at $62.98 on Aug 6 was sold at $63.12 the next day: a textbook, if tiny, pullback capture. Less flattering was CAT, bought at $887.13 on Aug 4 and sold at $843.72 on Aug 7 — a roughly 5% haircut that shows the strategy's exits don't always beat the drawdown. An earlier PG round-trip (sold $150.90, then re-bought $144.98) fit the playbook cleanly. Total equity has hovered near $9.9K–$10.0K, so the recent tape is essentially flat.
Backtest and validation
The headline backtest is respectable but unremarkable: +7.62% total return over 451 days, a 4.19% CAGR, and a 14.83% max drawdown. The Sharpe of 0.40 is thin, and the win rate of 39.39% across 137 trades confirms this is a strategy that loses more often than it wins and relies on winners outrunning losers. Turnover above 2,300% is high — this thing trades a lot relative to its edge, and fees ($137 in the test) are a real, if modest, drag.
The more important story is that walk-forward validation did not pass. Three of four folds were positive, and the out-of-sample slice (fold 4) returned +3.63% at a 0.74 Sharpe — encouraging. But fold 2 (Jan–Jul 2025) lost 11.42% at a punishing -1.7 Sharpe, exposing a regime where fading pullbacks simply failed. Most telling: with 7 trials in the search, the probabilistic Sharpe (PSR) sits at a healthy 0.70, but the deflated Sharpe (DSR) collapses to 0.196 once you adjust for multiple testing. That gap is the overfitting tax, and it's why the gate said no.
Verdict
Channel-pullback is a plausible, disciplined idea that is currently live but unproven. Its strengths are clean logic, healthy execution, and a genuinely strong recent fold. Its risks are a fragile edge (low Sharpe, sub-40% win rate), demonstrated regime sensitivity, and a deflated Sharpe that argues the backtest flatters it. Keep it on paper-sized capital and watch whether the out-of-sample momentum holds before trusting the thesis with more.