The thesis
news-sentiment runs a simple, intuitive rule: buy when recent news sentiment turns positive, and exit when it turns negative. It operates over a 24-name large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), and consumer/industrial staples (PG, KO, WMT, COST, CAT, XOM). The strategy is currently live.
The appeal is clear: sentiment is a fast-moving signal, and gating entries on positive news while cutting on negative news is a disciplined, one-directional bet. The catch is that a good rule only helps when it actually fires.
Recent activity
Live execution has been quiet to the point of dormant. Across scheduled runs from July 10 through July 17, the strategy executed zero trades and logged several rejected orders (one each on July 13, 14, and 15). Cash has sat unchanged at $990.06 while total portfolio value drifted with the market between roughly $9,990 and $10,251 — movement driven by held positions, not new decisions.
The last executed trades were all buys, clustered in late May through early July: AAPL (6 sh @ $312.06), MSFT (4 @ $428.23), BAC (35 @ $55.93), UNH (4 @ $402.85), and GOOGL (5 @ $360.40). No sells appear in the recent record, consistent with a signal that is currently accumulating rather than rotating.
Backtest and validation
Here the picture is sobering. Over 451 days the backtest returned just 0.19% (final equity $10,018.55, CAGR ~0.1%) on only 2 trades, with a 0% win rate and a Sharpe of 0.74. Maximum drawdown was a trivial 0.05% — but that is a direct consequence of near-total inactivity, not skill at loss avoidance. Turnover was 36.53% and total fees just $2.
Cross-validation did not pass. Of four walk-forward folds, three (Aug 2024–Dec 2025) produced zero trades and zero return; the entire 0.19% out-of-sample result comes from fold 4 alone, which delivered 1.5 Sharpe on its 2 trades. The deflated statistics tell the story: a strong PSR of 0.923 drops to a DSR of 0.551 once six trials are accounted for. One good window is not evidence of a repeatable edge.
Strengths and risks
Strengths: the logic is transparent, capital-preserving, and cheap to run; the single active fold hints the signal can work when sentiment is decisive.
Risks: the signal almost never triggers, so the sample is far too thin to trust. A 0% backtest win rate, a failed validation, and returns concentrated in one fold all point to overfitting or an under-calibrated sentiment threshold. Recent live rejections suggest the entry gate may be too strict or misfiring.
Verdict: promising in premise, unproven in practice. Before scaling capital, the priority should be widening the sentiment trigger enough to generate a statistically meaningful trade count — then re-running validation.