The thesis
channel-pullback is a straightforward mean-reversion idea bolted onto a trend filter: buy pullbacks to the lower regression channel or volume support only in confirmed uptrends, then exit at the upper channel or resistance. It trades a 24-name universe of US large caps — the usual megacap tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), and defensives (JNJ, PG, KO). The logic is sound in principle: you are trying to buy dips inside things that are already going up, which is a well-worn edge. Whether the implementation captures it is another matter.
Recent live activity
The strategy is live and running on a daily schedule. Over the last week it executed a handful of small trades — most recently buying HD (6 shares at $312.34) and NVDA (9 at $212.39) on 2026-09-14, after rotating out of V, AAPL, and UNH on 2026-09-11. There was a notable round-trip in BAC: a 31-share sell at $62.48, a 30-share re-buy the next session at $63.04, then out again at $62.79 — churn that earns fees without earning much else.
Live equity has drifted in a narrow band, from roughly $9,519 down to $9,371 across the shown window. That is below the $10,761 the backtest reports as final equity, a reminder that paper history and live experience diverge.
Backtest and validation
The full backtest covers 451 days: +7.62% total return, 4.19% CAGR, Sharpe 0.40, max drawdown 14.83%, across 137 trades. Two numbers should give any reader pause. The win rate is 39.39% — this is a system that is wrong more often than it is right and depends on winners outrunning losers. And turnover is 2,311%, extraordinarily high for a 7.62% gross return; the edge is thin relative to the trading it generates.
Walk-forward validation is where the story sharpens. Three of four folds were positive, but the spread is violent: fold 2 lost 11.42% (Sharpe −1.70) while fold 3 gained 20.68% (Sharpe 3.86). Out-of-sample return averaged just 3.63% with an OOS Sharpe of 0.74. Crucially, our harness marks validation as not passed. The deflated Sharpe ratio — which penalises for the 7 trials run — sits at 0.196, well short of confidence, even though the probabilistic Sharpe (0.702) looks superficially okay.
The balanced view
Strengths: a coherent, explainable thesis; a liquid universe; and genuine positive out-of-sample performance in three of four windows. Risks: a sub-40% hit rate, punishing turnover, a −11% fold that shows the strategy can bleed badly in the wrong regime, and a DSR that says the headline return is not clearly distinguishable from luck after accounting for trials. This is a candidate worth keeping on paper and watching — not one to scale. The gate failed it for good reasons.