The thesis
channel-pullback is a mean-reversion overlay on trend. It waits for a confirmed uptrend, then buys dips into the lower regression channel or a volume-support zone, exiting into the upper channel or resistance. The universe is 24 large-cap, liquid names spanning tech, financials, healthcare, and staples — the kind of book where regression channels tend to behave and slippage stays low.
Recent activity
The strategy is live and trading a $10k paper account that has hovered near flat all week — total equity ranged from $10,010 to $10,085 across the last six scheduled runs, ending at $10,045.59 on July 29. Activity is steady but light: most runs execute zero to three orders with no rejections, and cash swings widely (from ~$2,679 to ~$8,365) as positions rotate.
The trade tape tells the pullback story cleanly. It has round-tripped PG twice — buying near $145–$146 and selling near $150.6–$150.9 — and worked UNH from the long side around $419–$420 with an exit at $433.57. It also bought KO at $81.44 and later trimmed at $88.05, and sold CAT into strength at $834.28. These are textbook buy-the-dip, sell-the-rip entries, and the recent PG and KO round-trips landed on the right side.
Backtest and validation
Over 451 days the backtest returned 7.62% (final equity $10,761.52, ~4.19% CAGR) with a Sharpe of just 0.40 and a 14.83% max drawdown. Win rate is a low 39.39% across 137 trades — meaning the edge, such as it is, comes from winners outrunning losers, not from being right often. Turnover is heavy at 2,311%, though fees stayed modest at $137.
The honest headline: validation failed. Walk-forward across four folds was positive in three of them, but inconsistently so. Fold 1 (+6.5%, Sharpe 1.25) and Fold 3 (+20.7%, Sharpe 3.86) were excellent; Fold 4 was a tepid +3.6%. The problem is Fold 2 (Jan–Jul 2025): −11.4% with a −1.7 Sharpe and a 16% drawdown. Out-of-sample return averaged 3.63% at a 0.74 Sharpe — respectable, but the deflated Sharpe ratio of 0.196 (against a PSR of 0.70 and 7 trials) is what tripped the gate. After adjusting for the number of trials, the measured edge is not distinguishable from luck.
Verdict
Strengths: a coherent, interpretable thesis, clean execution, and two strong out-of-sample windows. Risks: a sub-40% win rate, dependence on a few large winners, and a fold that lost double digits when the regime turned. It earns its live paper slot for continued observation, but the failed validation is a real flag — I would not promote capital here until Fold 2-style regimes are handled or the deflated Sharpe clears the bar.