The thesis
Dual-momentum is about as transparent as a trend strategy gets: hold the strongest trending names by 60-day return and exit on a trend break. It fishes in a 24-name pond of large-cap U.S. blue chips — the mega-cap tech complex (AAPL, MSFT, GOOGL, NVDA), payments (V, MA), healthcare (JNJ, UNH, PFE, ABBV), staples (PG, KO, WMT, COST), and a few cyclicals (CAT, XOM, CVX). The logic is sound and well-documented in the literature: let winners run, cut laggards. No forecasting, no cleverness — just ride what is already moving.
Backtest performance
The headline numbers are genuinely good. Over 451 days the strategy turned $10,000 into $12,349.63 — a 23.5% total return, a 12.52% CAGR, and a 0.95 Sharpe with a contained 15.67% max drawdown. That is a clean risk-adjusted profile for a long-only equity strategy.
Two caveats sit underneath it. First, the win rate is just 28.79% across 136 trades — fewer than one trade in three makes money. That is not a flaw; it is the signature of trend-following, where a handful of large winners carry the book while most positions are cut small. Second, turnover is extreme at 2,638%, and the $136 in fees is a direct drag that only compounds in a live, taxable, real-fee setting.
Validation: the gate said no
This is where honesty matters. The walk-forward validation failed. Three of four folds were positive, and the out-of-sample segment returned 13.34% at a strong 2.15 Sharpe — encouraging. But fold 2 (Jan–Jul 2025) lost 7.31% with a -1.05 Sharpe and a 17.15% drawdown, showing the strategy can bleed through a sideways or choppy trend regime.
The statistics tell the real story. The Probabilistic Sharpe Ratio is a healthy 0.893, but the Deflated Sharpe Ratio — which penalises for the 6 trials run — collapses to 0.476. After accounting for selection effort, the edge is not statistically convincing. The gate was right to reject it, and no amount of a pretty equity curve should override that.
Recent activity: nothing happening
The live tape is quiet to the point of silence. The last six scheduled runs (Sept 28 through Oct 5) each executed zero trades and rejected zero — the book is parked in $10,000 cash, fully uninvested. No name in the universe is clearing the trend filter. That is the strategy behaving as designed: when nothing qualifies, it holds cash rather than force a position. But it also means there is currently no live signal to validate the backtest against.
Verdict
Dual-momentum is a well-behaved, interpretable strategy with a strong backtest and a textbook risk profile — but a failed validation gate and a week of idle cash are the sober counterweights. Treat it as a candidate under observation, not a deployment. The next meaningful data point is the first live trade, whenever a trend finally breaks in its favour.