The thesis
Dual-momentum is about as clean as trend-following gets: rank a 24-name large-cap universe by 60-day return, hold the strongest names, and exit when the trend breaks. The universe spans megacap tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, WMT) and healthcare (JNJ, UNH) — enough sector breadth that the ranking, not stock-picking, drives allocation. It's a philosophy with decades of academic support, and its appeal is discipline: no discretionary calls, just relative strength in, weakness out.
Backtest performance
Over 451 trading days the strategy returned 23.5% (CAGR 12.52%), finishing at $12,349 on a $10k base, with a Sharpe of 0.95 and a max drawdown of 15.67%. Those are honest, middle-of-the-road numbers — respectable, not spectacular.
Two details deserve a hard look. First, the win rate is just 28.79% across 136 trades. That's not a red flag on its own — trend-following lives on asymmetric payoffs, where a few large winners carry many small losers — but it means the equity curve depends on the tails behaving. Second, turnover is 2,638%. The strategy churns its book many times over, and while total fees were modest here ($136), heavy turnover is exactly where slippage and real-world frictions bite hardest.
Validation: the strategy did not pass
This is the headline caveat. Walk-forward validation across four folds failed the gate, and the fold detail shows why. Three of four folds were positive, but fold 2 (Jan–Jul 2025) lost 7.31% with a −1.05 Sharpe and a 17.15% drawdown — the strategy's momentum signal got whipsawed in a choppy regime. Folds 3 and 4 were strong (25.59% and 13.34%, Sharpes of 3.32 and 2.15), which lifts the out-of-sample average to a 13.34% return at a 2.15 Sharpe.
The statistics are mixed rather than damning: PSR of 0.893 is encouraging, but the deflated Sharpe (DSR) of 0.476 — after accounting for 6 trials — sits right on the fence. In plain terms, once you penalise for how many variants were tried, the edge is not clearly distinguishable from luck. The failed gate is the system doing its job.
Recent activity: idle in cash
Since going live, dual-momentum has done nothing. The last six scheduled runs (Aug 12–19) each report 0 executed, 0 rejected, with the book flat at $10,000 in cash. Either no name currently clears the entry threshold or the trend filter is holding it out — defensible behaviour for a trend-follower in an indecisive tape, but it means there is no live track record yet to corroborate the backtest.
Verdict
Dual-momentum is a coherent, well-motivated strategy with a credible backtest and one genuinely strong out-of-sample stretch. But the failed validation, the fence-sitting DSR, the low win rate and the punishing turnover all argue for humility. It belongs on the watchlist, sized small, until live trades start filling in the gap between promise and proof.