Strategy Thesis
Channel-pullback targets confirmed uptrends across a 24-name universe of US large-caps — spanning technology, financials, healthcare, consumer staples, and energy. The rule is straightforward: buy when price retreats to the lower regression channel or a volume-identified support level, then exit at the upper channel or a defined resistance. It is a mean-reversion-within-trend approach, designed to harvest the elastic snap back after institutions absorb selling pressure.
Backtest Scorecard
Over 451 days (roughly 15 months) and 137 paper trades, the strategy returned +7.62% on a starting equity of $10,000, reaching a final equity of $10,761. Annualised, that is a 4.19% CAGR — below a passive index baseline over the same window. Risk-adjusted results are similarly uninspiring: the full-period Sharpe sits at 0.40, and maximum drawdown reached 14.83% — a meaningful haircut for a strategy targeting high-conviction, low-volatility entries.
The win rate of 39.4% (54 of 137 trades) is below the majority of entries, which means the strategy depends entirely on its winners outrunning its losers. Whether that payoff ratio is sustainable in live conditions remains an open question.
Validation: Not Yet Cleared
The formal cross-validation pass — four chronological folds, seven parameter trials — returned a fail. The Probabilistic Sharpe Ratio (PSR) of 0.70 falls short of a conventional significance threshold, and the Deflated Sharpe Ratio (DSR) of 0.196 flags a high probability that the reported Sharpe is inflated by multiple-testing bias across the seven trial variants.
That said, the out-of-sample (OOS) fold — December 2025 through late May 2026 — delivered +3.63% with a Sharpe of 0.74, which is actually the second-best risk-adjusted result of any fold. Three of four folds finished positive, which is an encouraging directional signal even if the magnitude fails the statistical bar.
Fold-by-Fold Breakdown
The variance across folds is the most instructive data point:
| Fold | Period | Return | Sharpe | Max DD |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +6.53% | 1.25 | 6.16% |
| 2 | Jan 2025 – Jul 2025 | −11.42% | −1.70 | 16.09% |
| 3 | Jul 2025 – Dec 2025 | +20.68% | 3.86 | 3.26% |
| 4 | Dec 2025 – May 2026 | +3.63% | 0.74 | 8.44% |
Fold 3 is a standout — nearly 21% in five months with a drawdown under 4% — but it elevates the aggregate return in a way that obscures Fold 2's sharp loss. The strategy appears to stall or reverse in trending-down or choppy regimes, which aligns with its design: it needs an intact uptrend to function.
Recent Activity — June 2026
The agent has been notably quiet in June. Most daily runs logged zero executions, with one or two signals actively rejected. The only executed buy this month was 7 shares of AAPL at $279.39 on June 25. Earlier in the month, a CAT round-trip (buy at $857.78, sell at $912.06 six days later) illustrates the thesis working cleanly — a pullback entry captured a 6.3% move to resistance.
Conversely, a MSFT position entered at $427.54 was closed at $400.31, a loss of roughly 6.3%, demonstrating the cost of entries that fail to snap back before the upper channel becomes irrelevant.
Portfolio stands at approximately $10,041 total with $1,867 in cash (~18.6% idle), suggesting the agent is holding out for cleaner setups rather than forcing entries.
Risks to Watch
The low DSR is the headline concern: with seven parameter trials, the risk of having selected a historically lucky configuration is material. The strategy's utility in non-trending markets remains unproven given only one fold captured a sustained drawdown. High turnover (catalogued across 137 trades in 15 months) also means that in a live, fee-bearing environment, net returns could compress meaningfully beyond the $137 flat-fee assumption used in backtesting.