The thesis
momentum-code runs a simple, legible idea: buy the top positive movers in its universe, capped per position. The universe is 24 large-cap U.S. names spanning tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), staples (PG, KO, WMT), healthcare (JNJ, UNH, PFE), and energy (XOM, CVX). It is a classic cross-sectional momentum tilt with position caps as the only risk brake. There is no explicit sell rule in the data provided, which matters for how we read its numbers.
Backtest performance
Over 451 days the strategy returned 19.76% (final equity $11,976 on a $10k base), a 10.6% CAGR. That headline is respectable, but the shape is less flattering: a Sharpe of 0.65 against a maximum drawdown of 20.48%. In other words, the worst peak-to-trough loss was roughly as large as the entire realized return — a thin risk-adjusted cushion. Turnover of 92.74% on just 5 trades confirms this is a concentrated, low-diversification book. The reported win rate of 0% is a quirk worth flagging: with five buys and no closes in the record, there are no round-trips to score, so treat that figure as "not applicable" rather than a genuine 0-for-5.
Validation: good folds, failed gate
Here the story gets interesting. Walk-forward validation ran 4 folds, all 4 positive (+15.38%, +3.27%, +12.18%, +13.71%), with out-of-sample Sharpes as high as 2.47 and a most-recent OOS fold of +13.71% at Sharpe 1.93. On the surface that looks stronger than the full-sample Sharpe of 0.65.
And yet the gate reads passed: false. The reason is in the deflation stats. The Probabilistic Sharpe Ratio is a healthy 0.811, but the Deflated Sharpe Ratio is only 0.338 — computed across 6 trials. The DSR penalizes the Sharpe for multiple-testing selection bias, and 0.338 is well short of a confident deployment threshold. The system is correctly refusing to over-credit a result that may owe something to trial selection. This is exactly the guardrail you want: four pretty folds are not enough when six variants were fished from the same data.
Live activity: a book that has stopped
The live account tells a third story. The last executed trades were the initial buys on May 31–June 1 (MSFT, HON, BAC, then NVDA and XOM). Every scheduled run since July 17 through July 24 shows 0 executed with 1–3 signals rejected, against a static cash balance of $608.79. The strategy is fully invested and simply cannot afford its own signals; total value has drifted between roughly $9,870 and $10,130 while it holds.
Verdict
momentum-code is a clean, interpretable strategy with a genuinely encouraging fold history — but it is not validated for capital, and rightly so given a DSR of 0.34 and a drawdown that swallows its return. The live stall also exposes a design gap: no rebalancing or sizing logic to free cash for new signals. Strengths: consistency across folds, low complexity. Risks: shallow risk-adjusted returns, selection bias, and a book that has quietly gone dormant.