The thesis
channel-pullback is a mean-reversion strategy dressed in trend-following clothing. It waits for a confirmed uptrend, then buys pullbacks toward the lower regression channel or volume support, aiming to exit near the upper channel or resistance. The universe is a conservative 24-name large-cap basket — AAPL, MSFT, NVDA, JPM, V, JNJ, XOM and peers — so the edge is meant to come from timing entries and exits, not from exotic name selection.
Recent activity
The strategy is live and running on schedule. As of the 2026-09-25 run it held a total paper value of $11,037.49 with $3,930.01 in cash. Daily runs over the past week executed between zero and four trades, with a handful rejected — a sign the entry filters are doing their job rather than forcing trades.
The tape shows the mechanics working as designed. On 2026-09-21 it bought 4 MSFT at $498.37 and sold the same 4 shares on 2026-09-25 at $516.46 — a clean channel round-trip. It also rotated into energy (CVX at $204.99, XOM at $163.54) and NVDA (8 shares at $224.16) while trimming BAC, HD and PG. Not every trade is meaningful: a 1-share PFE buy at $27.90 and sell at $27.97 is noise, and hints at position-sizing that occasionally frays at the edges.
Backtest and validation
Over 451 days the backtest returned 7.62% (4.19% CAGR) with a Sharpe of 0.40 and a 14.83% max drawdown. The win rate is a low 39.39% across 137 trades — acceptable for a mean-reversion system if winners outrun losers, but the modest Sharpe says they only just do. Turnover is a striking 2,311%, and while total fees were only $137, that churn is a real drag on a thin edge.
The more important number is that validation failed. Three of four walk-forward folds were positive, and out-of-sample it still returned 3.63% at a respectable 0.74 Sharpe. But fold 2 (Jan–Jul 2025) lost 11.42% at a Sharpe of −1.70 — the strategy can clearly get caught leaning into a downtrend it mistook for a pullback. Critically, the deflated Sharpe ratio is just 0.196 against a probabilistic Sharpe of 0.702 across 7 trials. Once you penalise for the number of attempts, the confidence in a genuine edge nearly evaporates.
Verdict
channel-pullback is coherent, disciplined in execution, and modestly profitable on paper. But the fold-2 blowup and the near-zero deflated Sharpe mean the edge is fragile and regime-dependent. It earns a place in the live paper cohort as a learning experiment — not a candidate for meaningful capital until it demonstrates it can survive the next fold-2 without a double-digit drawdown.