The thesis
news-sentiment runs a simple, intuitive rule: buy names showing positive recent news sentiment and exit them when sentiment turns negative. It operates over a 24-name large-cap universe spanning mega-cap tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and energy. The strategy is currently live.
It is an appealing idea in principle — headlines move stocks, and a disciplined sentiment filter should catch momentum early and step aside before drawdowns. The question, as always, is whether the signal fires often enough, and cleanly enough, to matter.
Recent activity
The live log tells a telling story. Across scheduled runs from 18–25 September, the strategy executed trades on only one day: on 23 September it bought 4 shares of NVDA at $225.34 and sold 35 shares of BAC at $55.96. Every other session logged zero executions with one or two orders rejected — most likely a cash constraint, with reported cash sitting at just $990–$2,037 against a portfolio total hovering around $10,140–$10,340.
That BAC exit is instructive: the position was opened on 12 June at $55.93 and closed at $55.96 — a round trip that was, for practical purposes, flat. It captures the strategy's current character: active enough to trade, but not decisively enough to generate conviction-sized wins.
Backtest and validation
The headline backtest numbers are modest but not alarming: +0.19% total return, a 0.74 Sharpe, a very tight 0.05% max drawdown, and 36.5% turnover over 451 days, with only $2 in fees. Final equity was $10,018.55.
The caution flags sit in the detail. There were only 2 trades in the entire backtest, and the win rate is 0 — meaning the slim positive return rests on open or break-even positions rather than a track record of closing winners. That is a fragile foundation.
Walk-forward validation makes this explicit: it failed. Of four folds, only one (Dec 2025–May 2026) traded at all; the first three produced zero trades and zero return, so the signal simply went quiet across most of the test window. That single active fold looked healthy in isolation — 1.5 Sharpe on its 0.19% — and the probabilistic Sharpe ratio (PSR 0.923) is encouraging. But the deflated Sharpe ratio, which penalises for the 6 trials run, drops to 0.551, and one positive fold out of four is not evidence of robustness.
Verdict
The strengths are real: minimal drawdown, negligible fees, and a coherent, explainable thesis. The risks dominate for now. With a failed validation, a sample of two backtest trades, a zero win rate, and a live signal that rejects most of its own orders, news-sentiment reads as under-triggered rather than proven. The priority is not tuning returns but getting the signal to trade often enough — and with enough available cash — to produce a statistically meaningful record. Until then, treat it as a promising sketch, not a validated edge.