The thesis
Channel-pullback is a disciplined buy-the-dip overlay: it buys pullbacks to the lower regression channel or volume support only inside a confirmed uptrend, then exits at the upper channel or resistance. The "confirmed uptrend" filter is what separates this from naive dip-buying — the strategy is trying to catch mean reversion within a rising trend, not to catch falling knives. Its universe is 24 large-cap US names spanning tech, financials, healthcare, staples and energy, so it is diversified across sectors rather than betting on a single theme.
Recent activity
The book is running quietly. Of the last six scheduled runs (19–26 October), four executed zero trades, and total equity has drifted in a narrow band around $11,000 to $11,057 against roughly $7,391 of cash — a lot of dry powder sitting idle. The activity that did happen tells the story well: a clean NVDA round-trip (bought 9 shares at $220.45 on 14 Oct, sold at $220.92, re-bought at $211.74 on 22 Oct) shows the channel logic cycling in and out of the same name. October also saw exits in JNJ, PFE, KO, XOM and CVX, and a fresh AAPL entry (5 shares at $346.80). This is textbook pullback behaviour: frequent, small, two-sided.
Performance and validation
Here the picture turns cautious. The full backtest returns +7.62% over 451 days — a 4.19% CAGR — with a Sharpe of just 0.40 and a 14.83% max drawdown. The win rate is only 39.39% across 137 trades, which is not alarming on its own (pullback/trend systems routinely win less than half the time and rely on asymmetric winners), but the reward is not obviously paying for the risk. Turnover of 2,311% is very high, and while fees are modeled, that churn is a standing drag.
Most importantly, validation failed. Three of four walk-forward folds were positive, and some were excellent — fold 3 returned +20.68% at a 3.86 Sharpe — but fold 2 (Jan–Jul 2025) lost 11.42% at a −1.70 Sharpe. The out-of-sample return of 3.63% at a 0.74 Sharpe is respectable but thin. The decisive number is the deflated Sharpe ratio of 0.196: after adjusting for the 7 trials run during development, the evidence that this edge is real rather than lucky is weak. A PSR of 0.702 is suggestive but not conclusive.
The verdict
Channel-pullback is a coherent, well-diversified strategy that is making money live and behaving exactly as designed. But the honest read is that its edge is unproven: one bad regime wiped out a fold, headline Sharpe is low, and the deflated Sharpe says much of the backtest shine could be selection noise. It earns its keep as a low-conviction, risk-controlled sleeve — not as a strategy to size up until it survives more out-of-sample data.