Thesis
News-sentiment operates on a straightforward premise: accumulate positions when recent news flow for a stock turns positive, and exit when sentiment deteriorates. The strategy targets a 24-stock universe of large-cap US names spanning technology, financials, healthcare, consumer staples, and energy — a deliberately diversified canvas that should, in theory, keep the sentiment signal from being crowded out by sector-specific noise.
Recent Activity
The strategy has been running daily scheduled executions since late May, but it has been conspicuously quiet. Of the six most recent sessions (June 24–July 1), only one produced a trade: a 5-share buy of GOOGL at $360.40 on July 1. The prior five sessions saw zero executions despite the portfolio sitting on roughly $2,810 in idle cash.
Zooming out slightly, five trades have been placed since late May — AAPL, MSFT, BAC, UNH, and now GOOGL — spanning a mix of tech and diversified financials/healthcare. The portfolio stood at $9,861.89 total value as of July 1, with just $990 in cash remaining after the GOOGL entry. No exits have been triggered yet, which explains the reported 0% win rate: all open positions are unrealized.
Backtest & Validation Performance
Over 451 simulated days the strategy returned +0.19% with a Sharpe of 0.74, a maximum drawdown of just 5%, and only 2 completed trade round-trips. The CAGR of 0.10% is well below any reasonable benchmark, but the shallow drawdown profile is noteworthy for a sentiment-driven approach that could easily whipsaw.
The walk-forward validation tells a more sobering story. Across four equal folds spanning August 2024 to May 2026, only fold 4 generated any trades at all — folds 1 through 3 recorded zero executions, zero return, and zero drawdown. The strategy essentially sat in cash for roughly 16 months of the test window before coming alive in the December 2025–May 2026 fold, where it produced a +0.19% return and an impressive out-of-sample Sharpe of 1.50.
The validation check did not pass. With only 1 of 4 folds positive and the Deflated Sharpe Ratio (DSR) at 0.551 — well below the conventional 0.95 threshold given 6 total trials — the statistical case for genuine edge remains weak. The Probabilistic Sharpe Ratio (PSR) of 0.923 is more encouraging, suggesting the Sharpe estimate is probably positive, but it cannot compensate for such sparse signal generation.
Strengths
- Capital preservation: A 5% max drawdown over more than a year of live-equivalent exposure is genuinely attractive. The strategy does not blow up when it is idle.
- Fold-4 quality: The sole active period produced a 1.50 out-of-sample Sharpe — a high-quality signal when the model does fire.
- Simple, interpretable thesis: News-sentiment is easy to audit. When it makes a call, the reasoning can be traced directly to news headlines, which matters for debugging and trust.
Risks & Concerns
- Signal sparsity: A strategy that generates zero trades across three of four test periods is not providing the market exposure it promises. Sitting in cash is not a neutral outcome — it is opportunity cost.
- Regime dependency: All meaningful activity is concentrated in fold 4 (late 2025–mid 2026). This raises the possibility that the signal only functions in a specific news-cycle or volatility regime rather than across conditions.
- Failed validation gate: The strategy did not meet Headmars's walk-forward bar. Deploying larger notional capital or treating this as a primary signal would be premature.
- No closed trades on live paper book: Without any realized exits, the win rate and P&L picture are incomplete. The true test begins when positions are closed.
Outlook
News-sentiment shows a credible instinct — low drawdown, a clean signal when it fires, and a readable thesis. The immediate priority should be diagnosing why three full validation periods produced no activity: is the sentiment threshold too conservative, is the news-feed coverage too thin for most of the universe, or are there pipeline gaps? Adjusting signal sensitivity and re-running validation across a broader news corpus would be the logical next step before any capital commitment scales up.