The Thesis
momentum-code runs a classic idea: buy the top positive movers in its 24-name large-cap universe (AAPL, MSFT, NVDA, JPM, XOM, and peers), with a cap on each position. It is a trend-following bet — winners keep winning, at least for a while — constrained enough to avoid piling everything into a single hot ticker. The strategy is currently marked live.
Backtest Performance
Over a 451-day backtest, momentum-code returned 19.76%, lifting a $10,000 book to $11,976 for a CAGR of 10.6%. That headline number is respectable, but the risk profile deserves scrutiny. The Sharpe ratio is a modest 0.65, and the strategy sat through a 20.48% max drawdown — roughly the size of its entire cumulative gain. Turnover of 92.74% across just five trades, and a reported win rate of 0%, suggest the return came from a small number of concentrated, still-open positions rather than a broad base of realized winners. On this many trades, no single statistic should be trusted too far.
Validation: Encouraging Folds, Failing Grade
Here the picture is genuinely mixed. Across four walk-forward folds, all four were positive (15.38%, 3.27%, 12.18%, 13.71%), and the out-of-sample slice posted a strong 13.71% return at a 1.93 Sharpe. Fold-level Sharpes reached 2.47 and 2.31 with drawdowns near 6% — much cleaner than the full-sample 20% figure.
And yet validation did not pass. The deflated Sharpe ratio (DSR) sits at just 0.338, dragged down by six trials of search, even as the probabilistic Sharpe (PSR) reads 0.811. In plain terms: the folds look good, but after accounting for how many variants were tried, the framework isn't confident the edge is real rather than lucky. That is the validation layer doing exactly its job.
Live Activity: A Strategy That Stopped Trading
The most concerning signal is operational. After a burst of five executed buys (XOM, NVDA, BAC, HON, MSFT) around May 31–June 1, every scheduled run since July 10 has executed zero orders, rejecting one or two candidates each day. Cash is pinned at $608.79 — almost certainly too little to open a new capped position — and total account value has drifted between roughly $9,720 and $9,920, currently $9,873 (2026-07-17), sitting below its $10,000 starting line.
Verdict
momentum-code has a coherent thesis and genuinely encouraging fold-by-fold results. But the failed validation gate, the thin trade count, and a fully-invested book that can no longer act should temper enthusiasm. This is a strategy to watch — not yet one to trust with fresh capital.