Thesis
momentum-code runs one of the oldest ideas in systematic trading: buy the top positive movers in a defined universe and cap each position to limit concentration. Its universe is 24 US large caps spanning tech, financials, healthcare, staples, and energy — AAPL, MSFT, NVDA, JPM, V, JNJ, XOM, and peers. There is no shorting and no market-timing overlay; the strategy simply rotates into whatever is leading and trims exposure per name.
Backtest and Validation
Over a 451-day backtest, momentum-code returned 19.76% (10.6% CAGR), ending at $11,976 on a $10,000 base with just $5 in fees and turnover of 92.7%. The headline Sharpe is a modest 0.65, and the maximum drawdown of 20.48% nearly equals the total return — a reminder that the equity curve was earned through real volatility, not a smooth ride.
The walk-forward results are more encouraging. Across four folds, every single one was positive: +15.38%, +3.27%, +12.18%, and +13.71%, with fold Sharpes ranging from 0.46 to 2.47. Out-of-sample return came in at 13.71% with an OOS Sharpe of 1.93. The Probabilistic Sharpe Ratio is a healthy 0.811.
And yet validation failed. The culprit is the Deflated Sharpe Ratio of 0.338 — once the metric is penalized for six trials of selection, the edge no longer clears the bar. There is also a telling gap between the full-period Sharpe (0.65) and the much higher per-fold figures, which points to losses concentrated around fold boundaries that the segmented view flatters over. Four-for-four positive folds is genuinely good; failing the deflated test is a healthy dose of skepticism about how much of that is signal.
Recent Activity
Here the picture turns awkward. The last executed trades were on May 31–June 1, 2026 — buys of MSFT, HON, BAC, NVDA, and XOM. Since then, every scheduled run has done nothing: the last six sessions (Aug 6–13) each logged 0 executed with one or two orders rejected. Cash has sat frozen at $608.79 the entire time, and the account total has drifted between roughly $10,682 and $10,828 purely on mark-to-market moves in the existing holdings.
The rejections against a near-empty cash balance strongly suggest the strategy wants to buy but cannot fund new positions — it is effectively fully invested and idle. Note also that the reported win rate is 0 across 5 trades, consistent with positions that are still open rather than round-tripped.
Verdict
momentum-code has a defensible thesis and a backtest that survives walk-forward segmentation — real strengths. But the failed deflated-Sharpe check, the drawdown-heavy full-period profile, and a live engine that has been rejecting orders for two months all argue for caution. The immediate operational question is not whether the edge is real, but why the strategy can no longer transact.