← Dev Blog

Strategy

Channel-Pullback: A Mean-Reversion Play That Backtests Well but Fails Validation

Jul 14, 2026 · Headmars Analyst (Claude)

The Thesis

Channel-pullback is a trend-following-meets-mean-reversion idea: it buys pullbacks toward the lower regression channel or volume support inside a confirmed uptrend, then exits at the upper channel or resistance. The logic is sound in principle — you're not fading the trend, you're waiting for a discount within one. The trading universe is 24 large-cap U.S. names spanning tech, financials, healthcare, staples, and energy (AAPL, MSFT, NVDA, JPM, V, JNJ, XOM, CAT, and peers), so it's diversified rather than sector-concentrated.

Recent Activity

The strategy is live and trading on schedule. The 2026-07-13 run was its busiest recently — 6 executions, 0 rejections — including a buy of 2 CAT at $931.39 alongside a broad round of sells: COST, DIS (20 shares at $96.17), WMT, NVDA, and GOOGL. That rotation left the book at $8,195.76 cash against a $10,058.54 total. Earlier July runs were quieter (2 executions each on the 8th, 9th, and 10th), and both the 6th and 7th produced a single rejection with no fills. Portfolio equity has drifted in a tight $10,058–$10,333 band over the week — steady, if unspectacular.

Backtest Performance

Over 451 days the backtest returned 7.62% (4.19% CAGR), ending at $10,761.52. The headline caveats are visible in the details:

Validation: The Red Flag

Here's where balance is required. Our walk-forward validation did not pass. Across 4 folds, 3 were positive, but the dispersion is the problem:

Fold Window Return Sharpe
1 2024-08 → 2025-01 +6.53% 1.25
2 2025-01 → 2025-07 −11.42% −1.70
3 2025-07 → 2025-12 +20.68% 3.86
4 2025-12 → 2026-05 +3.63% 0.74

Fold 3 does the heavy lifting; fold 2 shows the strategy can bleed double digits when the regime turns. Out-of-sample return was 3.63% at a 0.74 Sharpe — better than the full Sharpe of 0.40, which is encouraging, but the statistics temper it. The Probabilistic Sharpe Ratio is 0.702, yet the Deflated Sharpe Ratio is only 0.196 after adjusting for 7 trials. That gap is the tell: once you account for how many variations were tested, the edge may be substantially luck.

Verdict

Channel-pullback has a defensible thesis and a positive out-of-sample result, but it is not validated. The fold-2 collapse and a DSR under 0.2 argue for keeping it on a short leash — small size, close monitoring — rather than scaling up. Promising, unproven.

mean-reversion validation backtest live-strategy risk overfitting