Thesis
news-sentiment operates on a disciplined premise: scan the news feed across a 24-stock universe of large-cap US equities, buy when aggregate sentiment turns positive, and exit when it sours. The universe spans six sectors — technology, financials, healthcare, consumer staples, consumer discretionary, energy, and industrials — offering breadth while staying within liquid, well-covered names.
Recent Activity
The strategy has been largely quiet heading into late June. Of six scheduled daily runs between 16–23 June 2026, only one produced an execution: a four-share purchase of UNH at $402.85 on 18 June. The portfolio now carries positions in AAPL (6 shares at $312.06), MSFT (4 shares at $428.23), BAC (35 shares at $55.93), and UNH, with $2,810.08 in cash against a total portfolio value of approximately $9,748.
The low trade frequency is by design: the strategy only acts when sentiment crosses a meaningful positive threshold. That threshold has been met on four occasions since late May — all buys, no exits recorded.
Backtest Performance
Over 451 backtest days, news-sentiment reached a final equity of $10,018.55 on a total return of 0.19% (annualised CAGR of roughly 0.10%). The Sharpe ratio is 0.74, and maximum drawdown is a minimal 0.05% — the strategy has not materially lost capital.
Two important caveats temper those numbers. First, only two trades underlie the backtest, which means the figures carry negligible statistical weight. Second, the reported win rate of 0% appears to reflect open, unrealised positions rather than closed losses: no sell-side exits appear in the data, so no completed round-trips exist to score.
Turnover of 36.53% hints at the strategy's rotation potential when conditions align, but the gap between that figure and actual trade count suggests backtest activity was concentrated in a very narrow window.
Validation
The four-fold cross-validation result is the clearest concern in this update: the strategy did not pass validation. Three of four folds — covering August 2024 through December 2025 — registered zero trades, zero return, and zero drawdown. Every recorded activity falls inside fold 4 (December 2025 – May 2026), which produced a +0.19% return and a Sharpe of 1.5.
The probabilistic Sharpe ratio (PSR) of 0.923 provides a degree of statistical encouragement — there is a 92.3% probability the Sharpe is above zero. The deflated Sharpe ratio (DSR) of 0.551, which penalises across six trials, is less reassuring.
Strengths
- Capital preservation: near-zero drawdown suggests the strategy avoids costly false positives.
- Fold-4 out-of-sample signal: when conditions aligned, the OOS Sharpe of 1.5 is a meaningful data point.
- Broad, liquid universe: 24 blue-chip names limit single-stock event risk and ensure fills.
Risks
- Signal sparsity: three dormant folds across sixteen months raise serious questions about whether positive-sentiment triggers materialise reliably across varying market regimes.
- Open-position bias: all trades to date are buys; without observed exits, the net-return and win-rate picture remains structurally incomplete.
- Regime concentration: all backtest performance is tied to a single period, suggesting the edge may be regime-specific rather than durable.
Outlook
news-sentiment is a patient, low-noise strategy that has not yet demonstrated consistent cross-regime reliability. The recent entries in UNH and BAC confirm the signal generator is functioning, but the strategy needs closed-trade data and fold coverage across a wider range of conditions before its edge can be considered validated. Worth monitoring — not yet ready to scale.