The Thesis
momentum-code runs one of the oldest ideas in quant investing: buy the strongest positive movers in the universe and cap each position so no single name dominates. Its hunting ground is a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and energy. The logic is deliberately simple — no forecasting, just riding whatever is already trending up.
Backtest Performance
Over 451 days the strategy returned 19.76%, ending at $11,976 on a $10k base for a 10.6% CAGR. The headline number is respectable, but the risk profile is less flattering: a Sharpe of 0.65 and a 20.48% max drawdown mean returns came with meaningful volatility — roughly one unit of drawdown for every unit of gain. Turnover of 92.7% is moderate for a momentum book, and fees were negligible ($5 total). One data quirk worth flagging: the reported win rate is 0 across just 5 trades — the positions appear to be open, unrealized buys rather than closed round-trips, so the metric is uninformative here rather than genuinely alarming.
Validation: The Interesting Part
The walk-forward validation is where this strategy earns a second look. Across 4 folds, all 4 were positive (15.38%, 3.27%, 12.18%, 13.71%), and out-of-sample it delivered a 13.71% return at a 1.93 Sharpe — notably better risk-adjusted performance than the full-sample 0.65. Fold-level Sharpes as high as 2.47 with single-digit drawdowns suggest the edge held up outside the training window.
And yet validation.passed is false. The deflated Sharpe ratio (DSR) sits at just 0.338, and with 6 trials run, the gate is correctly penalizing for selection bias — the probabilistic Sharpe (0.811) is decent but not enough to clear the multiple-testing hurdle. This is the system working as intended: strong-looking out-of-sample results are exactly what overfitting produces, and refusing to promote on 5 trades is prudent.
Recent Activity: Stuck in Neutral
Live, the strategy is quiet to the point of dormant. Its last executions were a cluster of buys in late May and early June — XOM, NVDA, BAC, HON, MSFT. Since then, every scheduled run from July 23–30 executed zero trades, rejecting 1–3 candidates each time. Cash has been pinned at $608.79 the entire week while total equity drifted between ~$10,085 and ~$10,292 (a modest ~2–3% above its $10k start). The pattern points to a capital constraint: the book is fully invested and can't add to the movers it wants without freeing cash.
The Balanced View
Strengths: a coherent, transparent thesis; uniformly positive folds; and out-of-sample risk-adjusted numbers that beat the full sample. Risks: a 20% drawdown, a failed statistical gate on a thin trade count, and a live book that has been unable to act for over a week. momentum-code is a promising candidate — but the validation layer is right to keep it on probation until more trades and a higher DSR confirm the edge is real, not a lucky draw from six trials.