The thesis
The news-sentiment strategy runs a simple, intuitive rule across a 24-name large-cap universe (AAPL, MSFT, NVDA, JPM, JNJ, XOM and peers): buy on positive recent news sentiment, exit on negative. It is a well-worn idea — sentiment as a short-horizon edge — and its appeal is that the logic is legible. The problem, as the data makes clear, is not the idea but how rarely it actually fires.
Recent activity
The live paper book is thinly active. Across scheduled runs from 21–28 September, the strategy executed trades on only one day (23 September: 2 executed, 1 rejected) and rejected every candidate on the other five. That 23 September session bought 4 shares of NVDA at $225.34 and sold 35 shares of BAC at $55.96. Earlier executed buys include GOOGL, UNH, MSFT and AAPL spread across June and July. The recurring pattern of 0 executed, 1–2 rejected is the headline story: signals are being generated but not clearing whatever entry filter sits behind them.
The live portfolio total has drifted in a narrow band — roughly $10,143 to $10,339 across the recent runs — sitting at $10,255.80 with $2,036.71 in cash as of 28 September. Steady, but not the mark of a strategy actively pressing an edge.
Backtest and validation
Over 451 days the backtest produced a 0.19% total return (CAGR ~0.1%), a Sharpe of 0.74, and a shallow 0.05% max drawdown — on just 2 trades with a reported 0% win rate and $2 in fees. That is effectively flat: capital preservation by inactivity rather than skill.
Crucially, walk-forward validation did not pass. Only 1 of 4 folds was positive, and that is generous framing — folds 1 through 3 logged zero trades. All performance came from fold 4 (Dec 2025–May 2026: +0.19%, Sharpe 1.5, 2 trades). The single-shot deflated metrics look superficially healthy (PSR 0.923), but the DSR of 0.551, adjusted for 6 trials, and the concentration of every result in one window tell the real story.
Strengths and risks
Strength: discipline. The strategy has not blown up, drawdown is negligible, and fees are trivial. Risk: there is almost no evidence it works. Three empty folds and two lifetime backtest trades mean we cannot distinguish edge from luck, and the persistent live rejections suggest the entry threshold may be miscalibrated for current news flow.
Verdict: promising premise, unproven execution. Before scaling, the priority is diagnosing why signals keep getting rejected and generating enough trades to validate on — not chasing the 0.19%.