Thesis
The news-sentiment strategy follows a deliberately simple rule: buy names with positive recent news sentiment, exit when sentiment turns negative. It trades a 24-name large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), and staples/industrials (PG, KO, WMT, COST, CAT, XOM). The logic is intuitive and the universe is liquid — exactly the kind of clean setup that should be easy to evaluate. The problem is that there isn't yet enough evidence to evaluate.
Recent activity
Live execution has stalled. Across six consecutive scheduled runs from 19 to 26 October, each report reads the same: 0 executed, 1 rejected. Cash has sat frozen at $2,036.71 the entire time, while total portfolio value has drifted down from $10,450.98 to $10,321.45 — a quiet erosion driven by the existing book, not by any new decision. The data doesn't say why the order keeps bouncing, but a signal that fires daily and never clears is a standing operational flag.
The last actual fills were back on 23 September (buy 4 NVDA @ $225.34, sell 35 BAC @ $55.96), preceded by a handful of 2026 entries in GOOGL, UNH, BAC, MSFT, and AAPL. So the portfolio is holding positions the model opened months ago while its current scheduled runs do nothing.
Backtest & validation
The headline backtest covers 451 days and returns a slim 0.19% (CAGR ~0.1%), ending at $10,018.55. Drawdown is a trivial 0.05% and fees total just $2 — but that's because the engine fired only 2 trades the entire window. Sharpe reads 0.74 and reported win rate is 0.
Walk-forward validation did not pass, and it's worth understanding why. Of four folds, three (Aug 2024–Dec 2025) produced zero trades and zero return. Every bit of performance — including the flattering OOS Sharpe of 1.5 and PSR of 0.923 — comes entirely from fold 4's two trades. With a DSR of 0.551 against 6 trials, the apparent out-of-sample strength is almost certainly an artifact of a single lucky window, not a repeatable edge.
Strengths and risks
Strengths: the thesis is coherent, costs are negligible, and drawdown has been minimal. There's no evidence of reckless behaviour here.
Risks — and they dominate: the strategy is effectively untested. Two backtest trades and one positive fold out of four is not a track record; it's noise. The failed validation, the win rate of 0, and the week of rejected live orders all point the same direction. Until the sentiment signal fires often enough to generate a real sample, the strong-looking PSR and OOS Sharpe should be read with heavy scepticism.
Verdict: a promising premise that has not yet earned conviction. Fix the rejected-order issue, then let it accumulate trades before trusting any of the performance numbers.