The thesis
Dual-momentum is a trend-rider. It holds the strongest-trending names in a 24-stock large-cap universe — ranked by 60-day return across the likes of AAPL, NVDA, JPM and XOM — and cuts a position the moment its trend breaks. The logic is classic time-series-plus-cross-sectional momentum: own what's working, exit fast when it stops. It's currently running live.
Recent activity
The more striking story is what the strategy hasn't done. Across six scheduled runs from 2 to 9 November, every single run executed 0 trades and rejected 0 trades, leaving the book at $10,000 cash, $10,000 total — fully flat. In other words, nothing in the universe is currently clearing the trend filter, so dual-momentum is parked in cash. That's the strategy behaving exactly as designed (no trend, no position), but it also means it is contributing no return right now, for better or worse.
Backtest performance
Over 451 trading days the backtest grew $10,000 into $12,349.63 — a 23.5% total return, or 12.52% CAGR. Supporting stats: Sharpe 0.95, max drawdown 15.67%, and 136 trades at $136 total fees with no FX cost.
Two numbers deserve a closer look. The win rate is just 28.79% — typical for trend-following, where a minority of large winners has to pay for a majority of small, quickly-cut losers. And turnover is 2,638%, confirming this is a high-churn approach that leans heavily on its exit discipline. A Sharpe under 1.0 on that much trading is respectable but not spectacular.
Validation: the harder test
Here's where enthusiasm needs tempering. Walk-forward validation failed (passed: false). Of four out-of-sample folds, three were positive — but fold 2 (Jan–Jul 2025) lost 7.31% with a 17.15% drawdown and a negative Sharpe of -1.05, a reminder of how badly momentum whipsaws in choppy, trendless regimes. Folds 3 and 4 recovered strongly (+25.59% and +13.34%), giving an out-of-sample return of 13.34% at an OOS Sharpe of 2.15.
The deciding metric is the Deflated Sharpe Ratio of 0.476. While the Probabilistic Sharpe Ratio sits at a comfortable 0.893, the DSR — which penalises for the 6 trials run during development — lands below 0.5, meaning we can't confidently reject the possibility that the edge is a product of selection rather than skill.
Strengths and risks
Strengths: a coherent, well-understood thesis; disciplined exits that kept max drawdown to a manageable ~16%; and three of four folds profitable with genuinely strong recent out-of-sample numbers.
Risks: the failed validation and sub-0.5 DSR are the headline caution — this edge may not survive multiple-testing scrutiny. The low win rate and high turnover make it fee- and regime-sensitive, and fold 2 shows the real pain a trendless market inflicts. For now, with the book in cash, patience is the position. We'd want to see it clear validation before treating the 23.5% as a repeatable edge rather than a favourable sample.