The thesis
Channel-pullback is a disciplined trend-following idea: buy pullbacks to the lower regression channel or a volume-support level inside a confirmed uptrend, then exit at the upper channel or overhead resistance. It trades a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, COST) and energy (XOM, CVX). The status is live, running on a scheduled paper account.
Recent activity
The last six scheduled runs (Oct 15–22) were quiet — mostly zero to two executions each. The standout is a near-flat round trip in NVDA: bought 9 shares at $220.45 on Oct 14, sold the same 9 at $220.92 on Oct 21, then re-entered 9 shares at a lower $211.74 the next day alongside 5 AAPL at $346.80. Earlier the book trimmed defensives and energy — selling JNJ, KO, XOM and CVX — and booked a small gain on PFE (bought $27.83, sold $28.17). The paper account total sits at $11,030.48 with $7,390.82 in cash as of Oct 22, so it is modestly ahead and not fully deployed.
Backtest and validation
Over 451 days and 137 trades the backtest returned 7.62% (CAGR 4.19%, final equity $10,761.52) with a Sharpe of 0.40 and a 14.83% max drawdown. Two numbers deserve attention. First, the win rate is only 39.39% — the strategy loses more often than it wins, so its edge depends entirely on winners being larger than losers, a classic asymmetric trend-following signature. Second, turnover is 2,311%, extremely high churn; fees stayed low here ($137, roughly $1/trade), but that sensitivity to transaction costs is a standing risk.
Validation did not pass. Across four walk-forward folds, three were positive, but the variance is brutal:
| Fold | Window | Return | Sharpe | Max DD |
|---|---|---|---|---|
| 1 | Aug'24–Jan'25 | +6.53% | 1.25 | 6.16% |
| 2 | Jan–Jul'25 | −11.42% | −1.70 | 16.09% |
| 3 | Jul–Dec'25 | +20.68% | 3.86 | 3.26% |
| 4 | Dec'25–May'26 | +3.63% | 0.74 | 8.44% |
Out-of-sample return is 3.63% at Sharpe 0.74 — respectable in isolation. But the Probabilistic Sharpe Ratio of 0.702 and especially the Deflated Sharpe of 0.196 (across 7 trials) tell the honest story: once you adjust for multiple testing, the confidence that the true Sharpe exceeds zero is weak.
The verdict
Strengths: a coherent, well-specified thesis; positive full-period and out-of-sample returns; and one genuinely strong regime (fold 3). Risks: a losing-majority win rate, a double-digit drawdown fold, punishing turnover, and a deflated Sharpe that fails the overfitting gate. Channel-pullback looks like a strategy whose edge is real but regime-dependent — worth keeping live and observed, not worth sizing up until out-of-sample results survive the deflation test.