The thesis
News-sentiment runs a simple, intuitive rule across a 24-name large-cap universe (AAPL, MSFT, NVDA, JPM, JNJ, XOM and peers): buy when recent news sentiment turns positive, exit when it turns negative. It's a clean idea with obvious appeal — headlines move prices, and a systematic reader that acts faster than a human could, in theory, capture that drift. The strategy is currently flagged live.
Recent activity
The live tape tells a quieter story than the thesis promises. Across scheduled runs from September 9 through 16, the strategy executed zero trades while rejecting one or two candidate orders on most days. Cash has been frozen at $990.06 the entire week, and portfolio total drifted down from roughly $10,377 (Sep 11) to $10,210 (Sep 16) — a passive mark-to-market decline on positions already held, not the result of active decisions.
The last actual fills came earlier, in June and July: executed buys in AAPL, MSFT, BAC, UNH and GOOGL. Since then the engine has been generating signals that its own risk or cash gates keep rejecting. That's not necessarily a fault — a strategy declining to over-trade is often healthier than one forcing positions — but it does mean the live book is currently coasting rather than expressing the thesis.
Backtest and validation
Here is where caution is warranted. Over 451 backtested days the strategy returned just 0.19% (CAGR ~0.1%), on only 2 trades, with a Sharpe of 0.74, a max drawdown of ~0.05%, turnover of 36.5% and $2 in fees. The win rate reads 0%. In practice this is a near-flat, barely-active result: with two trades over more than a year, there simply isn't enough activity to distinguish skill from noise.
The walk-forward validation makes that explicit — it did not pass. Of four folds, only one was positive, and critically, three of the four folds produced zero trades at all. The entire result rests on fold 4 (Dec 2025–May 2026), which did 2 trades for a 0.19% gain and a 1.5 Sharpe. A probabilistic Sharpe ratio of 0.923 looks encouraging in isolation, but the deflated Sharpe ratio — which penalizes for the 6 trials run — falls to 0.551, roughly a coin-flip after accounting for selection.
Verdict
Strengths: low drawdown, negligible fees, no FX cost, and a strategy that refuses to force trades it doesn't like. The idea is sound and the risk footprint is tiny.
Risks: the evidence base is dangerously thin. A single active fold, a failed validation, a zero win rate, and a live book that hasn't fired in months mean we have almost no data confirming the sentiment signal actually works. This one belongs in observation, not conviction — it needs to trade meaningfully across more regimes before anyone should trust the edge is real.