The thesis
momentum-code runs a deliberately simple idea: within a fixed 24-name large-cap universe — the usual megacap suspects across tech, financials, healthcare, staples and energy — buy the strongest positive movers and cap the size of any single position. There's no exotic signal here, which is a feature: it's easy to reason about and hard to overfit in the usual curve-fitting sense.
Recent activity
The last executed trades trace back to late May and early June 2026 — buys in MSFT, HON and BAC on 31 May, then XOM and NVDA on 1 June. Since then the strategy has gone quiet. Six scheduled runs between 9 and 16 September each executed zero trades while rejecting one to three candidates. With cash of just $608.79 against a portfolio of roughly $10,300, the pattern is consistent with a fully-invested book bumping up against its per-position caps and cash floor rather than a lack of signals. Portfolio value has drifted in a narrow $10,266–$10,541 band across that window.
Backtest and validation
Over 451 days the backtest returned 19.76% (10.6% CAGR) to a final equity of $11,976, with a Sharpe of 0.65 and a max drawdown of 20.48%. Turnover was 92.74% and total fees a trivial $5. Walk-forward validation split the history into four folds, and all four were positive: 15.38%, 3.27%, 12.18% and 13.71%, with out-of-sample Sharpe reaching 1.93 on the most recent fold.
And yet the validation verdict is fail. The reason lives in the deflated statistics. After penalising for the six trials that were run, the deflated Sharpe ratio (DSR) comes in at just 0.338 — below the bar that separates skill from luck — even though the probabilistic Sharpe ratio (PSR) sits at a healthier 0.811. In plain terms: the strategy looks good, but not clearly good enough once you account for how many variants were tried.
Strengths
Four-for-four positive folds is genuine consistency, and two of those folds cleared a Sharpe above 2.3 with sub-7% drawdowns. Costs are negligible, and the transparent construction makes the results credible rather than suspiciously smooth.
Risks
Three flags stand out. First, the sample is thin — five trades over the full backtest is far too few to trust any higher-moment statistic, and the headline 0% win rate reflects that there were no closed round-trips to score, not a string of losers. Second, the 20.48% full-period drawdown exceeds any single fold's worst (16.55%), a reminder that stitched-together folds understate real peak-to-trough pain. Third, and most important, the failed validation gate is doing exactly its job: DSR 0.338 is a call for caution.
Verdict
momentum-code is a tidy, consistent concept with an honest failing grade. The folds are encouraging and the live book is holding its value, but the deflated metrics and sparse trade count argue against scaling it up. Watch it, don't lean on it — and let the validation gate stay red until more out-of-sample evidence accumulates.