The thesis
The news-sentiment strategy runs a simple, intuitive rule across a 24-name large-cap universe (AAPL, MSFT, GOOGL, NVDA, JPM, and peers): buy when recent news sentiment turns positive, exit when it turns negative. It's the kind of premise that reads well on paper — sentiment leads price, so ride the wave. The strategy is currently live, so it's worth asking whether the data supports that promotion.
Recent activity
Live execution has been quiet to the point of dormant. The scheduled runs from July 6–13 executed zero trades, with a couple of rejections along the way, while total portfolio value oscillated narrowly between roughly $9,957 and $10,073 on a fixed $990.06 cash balance. The positions currently on the book come from a cluster of buys earlier in the window: GOOGL (5 @ $360.40), UNH (4 @ $402.85), BAC (35 @ $55.93), MSFT (4 @ $428.23), and AAPL (6 @ $312.06), spanning late May to July 1. In short, the book was assembled weeks ago and has mostly sat still since — sentiment signals simply haven't been firing.
Backtest and validation
This is where enthusiasm should cool. Over 451 days the backtest produced a 0.19% total return ($10,018.55 final equity), a 0.10% CAGR, a Sharpe of 0.74, and a tiny 0.05% max drawdown — on just two trades with a reported 0% win rate. A near-flat, two-trade sample tells us almost nothing about edge; the small drawdown reflects inactivity, not skill. The 0% win rate against a marginally positive return also suggests positions may still be open or barely above water, and shouldn't be read as evidence of anything.
Cross-validation makes the case harder. The strategy failed validation: of four folds, only one was positive — fold 4 (Dec 2025–May 2026), which carried all the action with 2 trades, a 1.5 Sharpe, and the entire 0.19% return. Folds 1 through 3 did nothing at all. The out-of-sample numbers (0.19% return, 1.5 Sharpe) are just fold 4 wearing a different label.
The deflated-metrics picture is mixed. The probabilistic Sharpe ratio (PSR) of 0.923 looks encouraging, but the deflated Sharpe ratio (DSR) of 0.551 — adjusting for the 6 trials run — sits barely above a coin flip. When PSR and DSR diverge like this, the honest reading is: the raw result isn't clearly distinguishable from luck once you account for how many variants were tried.
Verdict
Strengths: the thesis is coherent, the universe is liquid, turnover is modest (36.5%), and fees are negligible ($2). Risks: the sample is far too thin to validate, performance concentrates entirely in one fold, validation failed outright, and the DSR undercuts the flattering PSR. Right now news-sentiment is a plausible idea with no statistical proof behind it. It deserves more trades and more time before its live status is taken as a verdict — not a demotion, but a firm "unproven."