What the Strategy Does
Channel-pullback looks for large-cap names — a 24-stock universe spanning tech, financials, healthcare, consumer, and energy — that are in confirmed uptrends but have pulled back to the lower bound of a linear regression channel or a volume-support level. Entries trigger at those inflection points; exits target the upper channel or a nearby resistance level.
The logic is sound. Regression channels are a widely-used trend-following filter, and buying volume-supported pullbacks inside an uptrend aligns with basic mean-reversion mechanics. The universe skews blue-chip and liquid, which keeps slippage assumptions realistic.
Recent Activity
The strategy is live and running daily at 19:00 UTC. The past week has been quiet — three consecutive sessions (June 5–9) saw zero executions, with one to two signals rejected each day, suggesting the universe is either stretched or not offering clean pullback setups at current levels.
The last active batch was June 4, when four trades executed: buys in DIS (1 share @ $99.29) and MSFT (4 shares @ $427.54), and exits from UNH (5 shares @ $396.30) and JPM (6 shares @ $311.52). Two days earlier on June 2, UNH was bought at $376.02 and CAT was sold at $908.35 — notably, UNH was re-entered two days after a prior exit, implying a fresh pullback signal formed quickly. Portfolio equity has drifted from $10,117 on June 2 to $10,213 as of June 9, a modest consolidation after that burst of activity.
Backtest Performance
| Metric | Value |
|---|---|
| Total return | 7.62% |
| CAGR | 4.19% |
| Sharpe ratio | 0.40 |
| Max drawdown | 14.83% |
| Win rate | 39.39% |
| Trades | 137 over 451 days |
A 39% win rate is not a red flag on its own — channel strategies often work by letting winners run further than losers. But a Sharpe of 0.40 is low, and a max drawdown of nearly 15% against a total return of 7.62% implies the strategy's risk-adjusted profile needs improvement.
Turnover of 2,311% annualised is striking. At $1/trade in fees (as modelled), the 137-trade cost is manageable in paper trading, but this level of churn would meaningfully erode live returns depending on commissions and spread.
Walk-Forward Validation: The Concern
The four-fold walk-forward tells a cautionary story:
| 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% |
Three of four folds are positive, which is encouraging. But the variance is extreme — Fold 2 saw a −11.4% drawdown period while Fold 3 delivered 20.7% at a Sharpe of 3.86. That inconsistency is a signal that the edge is highly regime-dependent. The out-of-sample return (Fold 4, the most recent period) is +3.63% with a Sharpe of 0.74 — respectable but subdued.
Most critically, the strategy failed validation. With a Probabilistic Sharpe Ratio of 0.702 and a Deflated Sharpe Ratio of 0.196 across 7 parameter trials, there is meaningful probability the observed Sharpe is noise rather than edge. The DSR in particular — which penalises for multiple testing — lands well below the conventional 0.50 threshold.
Verdict
Channel-pullback has a coherent thesis and real periods of strong performance, but the validation results counsel caution before scaling. The priority for this strategy should be reducing fold variance — either through tighter regime filters (e.g., only trade when broad-market trend is confirmed) or stricter drawdown-based position sizing. The recent low-activity stretch may simply reflect a non-trending market, which is actually the strategy behaving as designed.