The thesis
The news-sentiment strategy runs a deliberately simple rule: buy a name when its recent news sentiment turns positive, exit when it turns negative. It operates over a 24-symbol large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV) and staples/industrials (PG, KO, WMT, COST, CAT, HON, XOM, CVX). It is currently flagged live.
It is an appealing idea — sentiment is a fast-moving, intuitive signal — but the numbers suggest the implementation is far more cautious than the thesis implies.
Recent activity
The last six scheduled runs (20–27 Oct 2026) tell a consistent story: 0 executed, 1 rejected every single time. The strategy generated one candidate per run and declined all of them, holding cash of $2,036.71 and a total portfolio value drifting between $10,321 and $10,453. In short, it has been dormant for a week.
The last executed trades are older: an NVDA buy and a BAC sell on 23 Sep, preceded by buys in GOOGL, UNH, BAC, MSFT and AAPL between late May and early July. So the strategy does act — but in bursts, then goes quiet.
Backtest and validation
The backtest covers 451 days and ends at $10,018.55 in equity — a total return of 0.19% and a CAGR of roughly 0.1%. Sharpe is 0.74, max drawdown is a tiny 0.05%, turnover 36.53%, and total fees just $2 across only 2 trades. The reported win rate is 0, which sits oddly against a (marginally) positive return and is worth a data-integrity second look.
Validation did not pass. Across 4 walk-forward folds, only 1 was positive — and tellingly, folds 1 through 3 recorded zero trades and zero return. Every ounce of the strategy's performance comes from fold 4 (Dec 2025–May 2026): 2 trades, +0.19%, a fold-level Sharpe of 1.5. The headline out-of-sample Sharpe of 1.5 is therefore built on two trades in one window — statistically meaningless.
The probabilistic Sharpe (PSR 0.923) looks healthy, but the deflated Sharpe (DSR 0.551), which penalises the 6 trials run, is far more sober.
Verdict
Strengths: negligible drawdown, minimal fees, and no evidence of reckless churn. It isn't losing money.
Risks: this is a strategy with almost no sample. Three of four folds never traded, the full-period return is a rounding error, and the recent live cadence is all rejections. A Sharpe quoted from two trades is not a track record — it's noise wearing a suit. Until news-sentiment demonstrates it can act consistently and across multiple regimes, it should be treated as an unproven experiment, not a validated edge.