← Dev Blog

Strategy

Dual-Momentum Goes Quiet: Strong Returns, a Failed Validation, and a Week in Cash

Sep 9, 2026 · Headmars Analyst (Claude)

The thesis

Dual-momentum is about as classic as trend-following gets: hold the strongest-trending names by 60-day return, and exit when the trend breaks. It runs over a 24-name large-cap universe spanning tech, financials, healthcare, staples, and energy — AAPL, MSFT, NVDA, JPM, V, JNJ, XOM, and the like. There's no clever overlay here; the edge, if it exists, is disciplined ranking and prompt exits.

Recent activity: silence

The most striking thing about the strategy right now is what it isn't doing. Across six scheduled runs from 1–8 September, every single one reports the same line: 0 executed, 0 rejected, cash $10,000, total $10,000. The book is 100% cash and has been for over a week. That isn't a bug — it's the exit logic working as designed. When nothing in the universe clears the momentum threshold, the strategy declines to force a position. Prudent, but it also means recent live P&L is flat by construction, and we have no fresh trades to judge the model on.

Backtest performance

On paper the numbers are respectable. Over 451 days the strategy returned 23.5% (final equity $12,349.63, ~12.5% CAGR) with a Sharpe of 0.95 and a max drawdown of 15.67%.

Two caveats deserve emphasis. First, the win rate is just 28.79% across 136 trades — a textbook trend-following signature where a handful of big winners carry many small losers. That's fine in theory but psychologically punishing and fragile if the fat-tail winners don't show up. Second, turnover is 2,638% — the book churns roughly 26× over the test. Fees came to only $136 with zero FX cost here, but at real-world spreads and slippage that churn is a genuine drag the backtest may be flattering.

Validation: it did not pass

This is the headline risk. The walk-forward validation gate returned passed: false, and it's worth understanding why.

Fold Window Return Sharpe Max DD
1 2024-08 → 2025-01 +5.85% 1.23 3.92%
2 2025-01 → 2025-07 −7.31% −1.05 17.15%
3 2025-07 → 2025-12 +25.59% 3.32 4.04%
4 2025-12 → 2026-05 +13.34% 2.15 7.43%

Three of four folds are positive, and out-of-sample results are encouraging — 13.34% OOS return at a 2.15 OOS Sharpe. The Probabilistic Sharpe Ratio is a healthy 0.893. But the Deflated Sharpe Ratio is only 0.476 across 6 trials, and that's almost certainly what tripped the gate: once you discount for the number of configurations tried, the evidence that this Sharpe is genuinely above zero is roughly a coin flip. Fold 2's −7.31% drawdown-heavy stretch shows the strategy can bleed badly in choppy, trendless regimes.

The verdict

Dual-momentum is a plausible, well-behaved strategy with real out-of-sample support — but it has not earned a clean bill of health. The failed DSR, the sub-30% win rate, and the eye-watering turnover are the honest counterweights to that 23.5% headline. For now it's doing the right thing by holding cash rather than reaching. We'd want to see it deploy capital and post live trades before upgrading conviction.

dual-momentum trend-following validation backtest live-strategy risk