The thesis
Dual-momentum is about as classic as systematic equity strategies get: rank a universe of 24 large-cap US names by their trailing 60-day return, hold the strongest trends, and cut positions when the trend breaks. The universe is defensively broad — mega-cap tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, COST) and healthcare (JNJ, UNH) — so the model is expressing which trends to ride rather than making concentrated single-name bets. It is a momentum-follower, full stop, and it should be judged on how well it captures persistent trends while sidestepping reversals.
Backtest performance
Over 451 trading days the strategy returned 23.5%, ending at $12,349 on a $10k base — a CAGR of 12.52%. The Sharpe of 0.95 is solid without being spectacular, and the max drawdown of 15.67% is tolerable for a long-only equity book.
Two numbers deserve scrutiny. First, the 28.79% win rate: fewer than one trade in three is profitable. That is entirely normal for momentum — a handful of large winners carry the book while many small losses get cut — but it means the edge is fragile to anything that clips the fat right tail. Second, turnover of 2,638% across 136 trades. That is a lot of churning, and while modelled fees here are trivial ($136 total, no FX), real-world slippage on that cadence would erode the reported edge.
Validation: the strategy did not pass
This is the headline risk, and it is worth being blunt about: validation failed. Walk-forward testing across four folds shows the strategy is not robust. Three of four folds were positive, but the results are wildly uneven:
- Fold 1 (Aug 2024–Jan 2025): +5.85%, Sharpe 1.23
- Fold 2 (Jan–Jul 2025): −7.31%, Sharpe −1.05, drawdown 17.15%
- Fold 3 (Jul–Dec 2025): +25.59%, Sharpe 3.32
- Fold 4 (Dec 2025–May 2026): +13.34%, Sharpe 2.15
Almost the entire backtest return comes from one exceptional fold. The out-of-sample Sharpe of 2.15 looks flattering, but the Deflated Sharpe Ratio of 0.476 — adjusting for the six trials run — tells the real story: after accounting for how many variants were tested, the evidence for a genuine edge is thin. A PSR of 0.893 is encouraging, but the DSR is the number that governs the deployment gate, and it is well short.
Live activity: waiting in cash
The live book tells its own quiet story. Every scheduled run from 9–16 September executed zero trades — no entries, no exits — leaving the full $10,000 in cash. With no name clearing the trend threshold, the model is correctly refusing to force a position. That is disciplined behaviour, but it also means the strategy is currently earning nothing.
Verdict
Dual-momentum has a coherent thesis and a headline backtest that flatters it. But the low win rate, heavy turnover, single-fold dependence, and a failing DSR all point the same way: this is a strategy whose live edge is unproven. Watching it wait patiently in cash is reassuring; deploying real capital behind it would be premature.