The thesis
News-sentiment runs a simple, legible rule: buy names on positive recent news sentiment and exit when sentiment turns negative. It fishes in a 24-name large-cap pond spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), staples (PG, KO, WMT, COST), and a handful of industrials and energy names (CAT, HON, XOM, CVX). It is a clean, auditable idea — which is exactly why its thin trade record matters so much.
Live activity
The strategy is marked live, but lately it has been closer to idle. Its last executed orders were on 2026-09-23 — a 4-share NVDA buy at $225.34 and a 35-share BAC exit at $55.96 — alongside earlier executed positions in GOOGL, UNH, MSFT, and AAPL. Since then, the scheduled runs tell a quieter story: across 2026-10-15 through 10-22, the book executed nothing, logging a rejected order on four of those days. Cash sits flat at $2,036.71 and portfolio value has drifted in a narrow $10,345–$10,510 band. A signal engine that keeps getting orders rejected is a flag worth watching: the strategy may be seeing opportunities it cannot act on.
Backtest and validation
Here is where enthusiasm has to be tempered. Over 451 backtested days the strategy turned in a +0.19% total return (CAGR ~0.1%), a 0.74 Sharpe, and a very shallow 0.05% max drawdown — but on only 2 trades, a 0% win rate, and $2 in total fees. Final equity was $10,018.55. That is barely distinguishable from sitting in cash.
Validation did not pass. Of four walk-forward folds, three (covering Aug 2024 through Dec 2025) placed zero trades. The entire result comes from fold 4 (Dec 2025–May 2026): 2 trades, +0.19%, a flattering 1.5 Sharpe, 0.05% drawdown. So "1 of 4 folds positive" overstates things — three folds simply did nothing. The probabilistic Sharpe (0.923) looks healthy, but the deflated Sharpe (0.551), which penalizes the 6 trials run, is only marginal.
Strengths and risks
Strengths: the logic is transparent, drawdown has been negligible, costs are trivial, and the one active fold produced a respectable risk-adjusted number.
Risks, and they dominate: there is effectively no sample. Two trades and a 0% win rate cannot support a conclusion, and a strategy that is inactive across most of the test window has not demonstrated an edge so much as an absence of losses. The failed validation gate and the recent run of rejections both argue for keeping this one on a short leash — watch for the sentiment trigger to actually fire and for trades to clear before reading anything into the equity curve.