The thesis
The news-sentiment strategy runs a simple, intuitive rule: buy names carrying positive recent news sentiment, exit when sentiment turns negative. Its universe is a conservative slate of 24 US large-caps spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), staples, and energy. It is currently flagged live, but the data suggests it is live more in the sense of "switched on" than "pulling its weight."
Recent activity
The live paper account has been accumulating, not rotating. Over late May through early July it opened five positions — AAPL (6 @ $312.06), MSFT (4 @ $428.23), BAC (35 @ $55.93), UNH (4 @ $402.85), and GOOGL (5 @ $360.40) — and, tellingly, has recorded no exits. That matters, because the sell leg is half the thesis. Every scheduled run from July 21 to July 28 came back identical: 0 executed, 0 rejected, cash pinned at $990.06. Total account value drifted between roughly $9,879 and $10,193 purely on mark-to-market, not on decisions. The signal, right now, is silent.
Backtest and validation
The headline backtest is honest but underwhelming: +0.19% total return over 451 days, a 0.1% CAGR, Sharpe 0.74, and a barely-there 0.05% max drawdown, on just 2 trades and $2 in fees. A near-zero drawdown is easy to admire until you notice it comes from a strategy that almost never trades.
Validation is where the caution flags go up. The walk-forward suite did not pass. Of four folds, only one produced any activity at all: fold 4 (2025-12-16 → 2026-05-29) logged 2 trades, a +0.19% return, and a respectable 1.5 Sharpe. The other three folds sat at zero — no trades, no return, no information. So the entire out-of-sample record rests on a single window. The probabilistic Sharpe (PSR) reads a comfortable 0.923, but the deflated Sharpe (DSR) of 0.551 — which discounts for the 6 trials run — sits right on the coin-flip line. In plain terms: once you account for how many variants were tested, the edge is not statistically convincing.
The balance
Strengths: capital preservation is genuine — drawdowns are minimal, fees are trivial, and the one active fold behaved well (Sharpe 1.5). The universe is defensively diversified.
Risks: the sample is far too thin to trust — two backtest trades and a win rate that is effectively undefined (no completed winning round-trips). The signal fires so rarely that three of four folds never triggered, and the live book has bought without ever selling, leaving the exit rule unproven. With DSR near 0.5 and validation failing, this reads as a plausible idea awaiting evidence, not a deployable edge. I would keep it on a short leash: paper-only, and gated until the sentiment trigger demonstrates it can both enter and exit across more than one regime.