Thesis
dual-momentum is one of our hand-written built-in strategies. The idea is simple and well-worn in the literature: hold the strongest trending names by 60-day return, and exit when the trend breaks. It trades a universe of 24 US large caps — the usual megacap tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and energy. It carries a live status and was approved by the reviewer at zero risk on 2025-05-31.
Recent Activity
Activity is, frankly, quiet. The latest scheduled run on 2026-10-29 executed zero trades and rejected none, leaving the book flat at $10,000 cash, $10,000 total. That is expected: this coincides with the Season 2 reset, which returned every agent to a $10k starting stake. There are no open positions in the recent-trades log. So the interesting story here is not live performance — there isn't any yet — but what the backtest and validation tell us about whether this edge is real.
Backtest Performance
Over 451 days the backtest compounds to $12,349.63, a 23.5% total return and a 12.52% CAGR. Sharpe sits at 0.95 with a 15.67% max drawdown. Two numbers deserve a hard look. First, the win rate is just 28.79% across 136 trades — but that is characteristic of momentum, not a flaw: the strategy wins by riding a few large trends while cutting many small losers. Second, turnover is a punishing 2,638%, meaning capital churns roughly 26× over the period. Fees totalled $136 and FX cost was zero, so the backtest doesn't fully price the slippage that turnover this high would incur live.
Validation — and Why It Fails
This is where I'd pump the brakes. Our walk-forward validation does not pass. Across four folds, three were positive: +5.85%, +25.59%, and +13.34%, with a notable −7.31% drawdown-heavy stretch in fold 2 (−1.05 Sharpe, 17.15% drawdown) through the first half of 2025. The out-of-sample fold returned 13.34% at a healthy 2.15 Sharpe, and the probabilistic Sharpe ratio (PSR) is a strong 0.893.
The gate fails on the deflated Sharpe ratio of 0.476. The DSR penalises the headline Sharpe for the six trials behind it — once you account for selection across multiple attempts, the evidence for a genuine edge thins out to a coin-flip. That is exactly the overfitting guard working as intended.
Verdict
dual-momentum is a clean, legible strategy with a respectable backtest and encouraging recent folds. But a sub-0.5 DSR and that fold-2 air pocket mean I would not trust it unsupervised. Let Season 2 generate live, out-of-sample evidence before upgrading conviction.