The Thesis
Momentum-code runs one of the oldest ideas in systematic trading: buy the top positive movers in the universe, capped per position. The universe is a 24-name basket of US large caps spanning tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), staples (PG, KO, WMT, COST) and energy/industrials (XOM, CVX, CAT, HON). The per-position cap keeps the book from concentrating into a single winner — a sensible guardrail for a strategy whose entire edge is chasing strength.
Backtest and Validation
Over 451 days the backtest returned 19.76%, ending at $11,976 on a $10k base, for a 10.6% CAGR. That headline is respectable, but the risk-adjusted picture is more sober: a Sharpe of just 0.65 against a 20.48% max drawdown. Turnover ran at 92.74%, so this is not a low-churn strategy, though fees were negligible at $5.
The walk-forward analysis is where momentum-code earns real credibility. All four folds were positive — 15.38%, 3.27%, 12.18% and 13.71% — with out-of-sample Sharpes reaching 2.47 and 2.31 in the strongest windows. Consistency across regimes is exactly what separates a real edge from a curve-fit.
And yet the validation gate reports passed: false. The reason lives in the deflated statistics: the Probabilistic Sharpe Ratio sits at 0.811, but the Deflated Sharpe Ratio is only 0.338 once the six trials are penalized. In plain terms, the platform is discounting the result for the number of attempts it took to find it. Four green folds are encouraging; they are not yet enough to clear the anti-overfitting bar.
Recent Activity
Here the story cools. The last executed trades were a cluster of buys on May 31–June 1 — XOM, NVDA, BAC, HON and MSFT. Since then, the scheduled runs have gone quiet: from August 13 through 20, every run executed zero trades and rejected one or two, holding cash flat at $608.79. Over that same stretch the paper account slid from $10,804 to $10,568, a roughly 2% drawdown with no new positioning to respond to it.
One oddity worth flagging: the reported win rate is 0% across five trades. Given the positive equity, this almost certainly reflects unclosed round-trips rather than five losers — the positions are held, not realized — but it is a metric to watch as trades close out.
Verdict
Momentum-code has a clean, defensible thesis and unusually consistent out-of-sample behavior. The caution flags are real, though: a modest full-sample Sharpe, a 20% drawdown, a validation gate that has not cleared, and a live book that has stopped trading while quietly bleeding. This is a strategy worth keeping live on a short leash — promising enough to prove itself, unproven enough not to size up yet.