The thesis
The news-sentiment strategy runs on a deliberately simple premise: buy when recent news sentiment turns positive, and exit when it turns negative. It operates over a 24-name large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), and consumer staples and industrials (PG, KO, WMT, COST, CAT, XOM). It is currently flagged live.
It is a clean, legible idea — the kind that is easy to reason about and easy to audit. That clarity is a genuine strength. The question is whether the signal fires often enough, and cleanly enough, to earn its keep.
Recent activity
At the moment, the honest answer is: not much is happening. The six most recent scheduled runs — from 2026-09-29 through 2026-10-06 — each closed with 0 executed, 0 rejected. The paper account sits at a total of $10,366.91 against $2,036.71 in cash, so it is holding positions and drifting with the market rather than acting on fresh signals.
The last real decision came on 2026-09-23, when it bought 4 shares of NVDA at $225.34 and sold its 35-share BAC position at $55.96 — a near round-trip on the BAC it had bought back in June at $55.93. Earlier executed buys include GOOGL, UNH, MSFT, and AAPL across May through July. So the strategy does transact, but sparingly: a handful of trades over months, then long stretches of inactivity.
Backtest and validation
The backtest covers 451 days and returns +0.19% total (roughly 0.1% CAGR), with a Sharpe of 0.74, a very shallow 0.05% max drawdown, and just 2 trades at $2 in total fees. Capital preservation looks excellent; the return contribution is negligible, and the reported win rate of 0 alongside a positive total suggests the gains are sitting in open, marked-to-market positions rather than closed winners.
Validation is where the caution sharpens. The suite failed. Across four walk-forward folds, only one — fold 4 (2025-12-16 to 2026-05-29) — traded at all, contributing the entire 0.19% and a healthy 1.5 out-of-sample Sharpe. Folds 1 through 3 placed zero trades. A PSR of 0.923 looks reassuring, but the deflated Sharpe ratio of 0.551 across six trials, and the fact that three of four folds are empty, tell the real story: there simply isn't enough activity to trust the result.
Verdict
news-sentiment is low-risk and well-behaved, but statistically starved. One live fold and two backtest trades cannot distinguish skill from a quiet market. Before leaning on it, the signal needs to fire more often — or we need far more out-of-sample evidence that, when it does fire, it is right.