The thesis
momentum-code runs one of the oldest ideas in the book: buy the top positive movers in a fixed universe, capped per position. The universe is 24 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). It is currently flagged live.
Backtest and validation
Over 451 days the strategy returned 19.76%, lifting a $10k book to $11,976 — a 10.6% CAGR with a 20.48% max drawdown and turnover of 92.74%. Fees were a trivial $5 with no FX cost. The headline Sharpe is a modest 0.65.
The walk-forward picture is more flattering. All four folds were positive (15.38%, 3.27%, 12.18%, 13.71%), with out-of-sample Sharpes of 2.47, 0.46, 2.31 and 1.93. The most recent fold's 13.71% OOS return at a 1.93 Sharpe is genuinely encouraging, and per-fold drawdowns were shallow (5–6%) outside the mid-2025 soft patch where fold 2 dipped to a 0.46 Sharpe and a 16.55% drawdown.
So why does validation report passed: false? The culprit is the deflated Sharpe ratio: DSR 0.338, below the bar, even though the probabilistic Sharpe (PSR 0.811) looks healthy. With 6 trials in the search, the multiple-testing penalty bites, and the full-period Sharpe of 0.648 is simply too thin to survive deflation. The gate is doing its job — all-positive folds are reassuring, but they don't by themselves prove the edge isn't a product of selection.
A caveat on the numbers
Two figures deserve a skeptical eye. The reported win rate is 0% across 5 trades — not a losing streak, but an artifact: the five executed positions are still open, so there are no closed round-trips to score. Read "win rate" here as undefined, not catastrophic. The 20.48% full-period drawdown also dwarfs any single fold, a reminder that stitched sub-periods understate the pain of a continuous equity curve.
Recent activity
Here is the awkward part. The last executed trades were a cluster of buys back on 31 May–1 June 2026 — MSFT, HON, BAC, then NVDA and XOM. Since late September, every scheduled run tells the same story: 0 executed, 1–3 rejected, with cash frozen at $608.79 and total equity drifting around $10,300–$10,410. The strategy is effectively fully invested and cash-starved; new signals keep getting rejected because there is nothing to deploy. Live equity (~$10,410 on a $10k base) is up only ~4%, well short of the backtest.
Verdict
momentum-code has a coherent thesis and a believable out-of-sample track record, but it is not paper-perfect: the validation gate rejects it on a deflated Sharpe, and in production it has stalled into a hold-only posture for months. The watch items are clear — can it free cash to act on fresh momentum, and can the edge clear the deflation bar on more trials? Until then, treat it as promising but unconfirmed.