The thesis
News-sentiment is a long-only, large-cap strategy with a deliberately simple rule: buy names showing positive recent news sentiment, exit when sentiment turns negative. It trades a 24-name blue-chip universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), healthcare (JNJ, PFE, UNH) and staples (KO, PG, WMT). It is currently live.
Live activity
Since late October the strategy has deployed cash into five positions — V, NVDA and PFE on 29 Oct, KO on 3 Nov, and AAPL (5 shares @ $348.52) on 10 Nov. Notably, every execution has been a buy; no exits have fired, so the "sell on negative sentiment" half of the mandate remains untested in practice. Most scheduled runs do nothing at all: across the six sessions logged between 5 and 12 Nov, only the 10 Nov run executed a trade. Portfolio value has drifted lower over that window, from $9,758.90 (5 Nov) to $9,534.64 (12 Nov), with cash now at $626.15.
Backtest and validation
The backtest invites caution. Over 451 days it returned just +0.19% (final equity $10,018.55, ~0.1% CAGR) on only 2 trades, with a 0% win rate, a 0.74 Sharpe and a tiny 0.05% max drawdown. Fees were negligible ($2, no FX cost). A near-flat curve built on two trades sits uncomfortably close to noise.
Walk-forward validation did not pass. Only 1 of 4 folds was positive — and the other three executed zero trades, meaning the signal simply never fired across most of the test period. All the performance comes from fold 4 (Dec 2025–May 2026): +0.19%, Sharpe 1.5, 2 trades. The probabilistic Sharpe (PSR 0.923) looks flattering, but the deflated Sharpe (DSR 0.551) across 6 trials is far more sober and is the number to trust when an edge rests on so few observations.
Verdict
The appeal is genuine: shallow drawdowns, trivial costs and a clean, interpretable rule that is easy to reason about. But the evidence base is thin. Two backtest trades, a 0% win rate, three dead folds and a failed validation gate all point to a signal that triggers too rarely to prove itself either way. Live behaviour compounds the concern — the book is buying but never selling, so its downside discipline is unproven while equity quietly slips. News-sentiment is worth watching, not yet trusting: it needs to generate enough trades to clear validation on its own merits before any real conviction is warranted.