The thesis
News-sentiment runs a simple, intuitive idea: buy names showing positive recent news sentiment, exit when sentiment turns negative. It operates over a 24-stock large-cap universe spanning tech, financials, healthcare, staples, and energy — AAPL, MSFT, GOOGL, NVDA, JPM, V, JNJ, UNH, PG, KO, WMT, XOM, CVX, and more. It's a clean, explainable premise, and the strategy is currently flagged live.
The catch, as always with sentiment strategies, is that a good story needs a signal that actually fires often enough to be measured.
Recent activity
On the live paper book, news-sentiment did its buying earlier: five executed purchases between late May and early July — AAPL (6 sh @ $312.06), MSFT (4 @ $428.23), BAC (35 @ $55.93), UNH (4 @ $402.85), and GOOGL (5 @ $360.40). It is now near fully invested, holding just $990.06 in cash against a total book of roughly $10,215 as of 15 July.
The last week has been quiet by design. Scheduled runs on 13–15 July each executed 0 trades and rejected 1, and the prior three sessions executed nothing at all. No exits have printed, which is consistent with a reported win rate of 0 — no positions have been closed to book a result yet.
Backtest and validation
The backtest is where caution is warranted. Over 451 days, the strategy produced a 0.19% total return (≈0.1% CAGR), a Sharpe of 0.74, and a shallow 0.05% max drawdown — but on only 2 trades and $2 of fees, with turnover of 36.5%. Two trades over roughly a year and a half is not a track record; it's an anecdote.
Walk-forward validation makes the point sharper. The strategy failed, passing only 1 of 4 folds. Folds 1–3 (Aug 2024 through Dec 2025) recorded zero trades and zero return — the signal simply never triggered. All activity, and the entire 0.19% result, comes from fold 4 alone (Dec 2025–May 2026), which posted a Sharpe of 1.5 on its 2 trades. The Probabilistic Sharpe Ratio is a healthy 0.923, but the Deflated Sharpe Ratio of 0.551 — adjusted for 6 trials — sits barely above a coin flip.
Verdict: strengths and risks
Strengths: a legible thesis, disciplined risk (tiny drawdown, low fees), and no evidence of reckless churn. When it did trade, it didn't lose money.
Risks — and they dominate: the sample is far too small to be credible. Three of four validation windows never fired, and the headline number rests on a single fold and two trades. That is textbook fragility, not edge. Until news-sentiment demonstrates it can generate and close trades across multiple regimes, treat its numbers as unproven. The next milestone worth watching is the first exit signal — that's when this strategy finally gets a real data point.