The thesis
Channel-pullback is a trend-following mean-reversion hybrid: it buys pullbacks to the lower regression channel or volume support within confirmed uptrends, then exits at the upper channel or resistance. The idea is to pay a better entry price on names that are still structurally rising, rather than chasing breakouts. The universe is 24 liquid large caps spanning tech, financials, healthcare, staples and energy — AAPL, MSFT, NVDA, JPM, V, UNH, XOM and CAT among them — so the strategy leans on breadth rather than concentration.
Recent activity
The strategy is now live. Over the last week of scheduled 19:00 runs it has been actively rotating: it sold NVDA on three consecutive days (9 shares at $212.39, then $212.03, then $214.11) after buying the same block on 2026-09-14, alongside sells in V, AAPL and UNH and fresh buys in HD, BAC and XOM (11 shares at $163.54). Cash has swung from just $47.96 on 2026-09-10 — effectively fully invested — up to around $2,033 more recently, with total account value near $11,171 as of the 2026-09-17 run. Not every run fires: the 2026-09-17 session executed nothing and rejected two candidates, and other days show 1–3 rejections, consistent with a rules gate that declines setups failing its channel or trend filters.
Backtest and validation
Over 451 days the backtest returns +7.62% (final equity $10,761.52, ~4.19% CAGR) with a Sharpe of just 0.40 and a max drawdown of 14.83%. Win rate is a low 39.39% across 137 trades — the return comes from asymmetry, letting the resistance-target winners outrun the stopped losers, which is exactly what the thesis predicts. Turnover is aggressive at 2,311%, though fees stayed modest at $137 total and FX cost was zero.
The honest headline: validation did not pass. Walk-forward across four folds shows three positive, but fold 2 (2025-01-22 to 2025-07-05) lost -11.42% at a -1.70 Sharpe, while fold 3 posted a spectacular +20.68% at 3.86 Sharpe. That spread is regime dependence, not consistency. Out-of-sample return averaged 3.63% at a 0.74 Sharpe. Crucially, while the probabilistic Sharpe ratio (PSR 0.70) is respectable, the deflated Sharpe ratio is only 0.196 after adjusting for 7 trials — below the bar. In plain terms, once we penalise for how many variants were tried, the edge is not statistically convincing.
Verdict
Strengths: a coherent, well-diversified thesis and genuinely strong performance in trending windows. Risks: a sub-0.5 headline Sharpe, one deeply negative fold, and a failed deflated-Sharpe test that flags possible overfitting. Running it live is a reasonable way to gather forward evidence — but position sizing should stay conservative until out-of-sample results, not backtests, justify more conviction.