What the Strategy Does
The news-sentiment agent applies a straightforward rule: buy large-cap U.S. equities when recent news sentiment turns positive, exit when it turns negative. Its 24-symbol universe spans mega-cap tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), consumer staples and discretionary (PG, KO, WMT, COST, MCD, NKE, HD), energy (XOM, CVX), industrials (CAT, HON), and media (DIS). The thesis is that news flow is a leading — not lagging — signal for short-term price direction in liquid, widely-covered names.
Recent Activity
The strategy has been quiet. Over the past six scheduled daily runs (July 3–9), it executed zero trades. On July 7 it flagged one signal but rejected it, likely due to a risk or sizing constraint. Portfolio equity has ranged from $9,935 to $10,073 over that window, moving with mark-to-market on existing positions rather than new activity.
The most recent executions were two buys at the end of June: 5 shares of GOOGL at $360.40 on July 1, and 4 shares of UNH at $402.85 on June 18. Earlier in the live run, the agent accumulated MSFT (4 shares, $428.23) and AAPL (6 shares, $312.06). The current portfolio carries $990 in residual cash.
Backtest Performance
Over 451 days of backtesting, the strategy produced a +0.19% total return on a starting equity of ~$10,000, ending at $10,018.55. Annualized (CAGR), that is approximately +0.10% — below money-market rates for the period. The Sharpe ratio of 0.74 is sub-threshold by most institutional standards (>1.0 is typically the minimum bar). On the upside, max drawdown was a contained 5%, and total fees were just $2, reflecting very low turnover (36.5% annualized) and only 2 completed round-trips.
Win rate is reported as 0% — this reflects that no trades have been closed yet, not that all trades have lost money. The open positions are unrealized.
Validation: Where the Cracks Show
The four-fold walk-forward validation is the sharpest lens here, and it tells an uncomfortable story. Three of four folds produced zero trades and zero return. All activity is concentrated in Fold 4 (December 2025 – May 2026), which returned +0.19% with a Sharpe of 1.5. The other three folds — covering August 2024 through December 2025 — were completely inert.
The headline out-of-sample Sharpe of 1.5 looks strong, but it is derived entirely from a single fold with only 2 trades. The Probabilistic Sharpe Ratio (PSR) of 0.923 suggests the Sharpe is likely above zero, which is something — but the Deflated Sharpe Ratio (DSR) of 0.551 adjusts for the fact that 6 parameter trials were run, and it drops the confidence substantially. The strategy has not passed Headmars's internal validation gate.
Strengths and Risks
Strengths: Low drawdown, minimal fees, and a clean thesis grounded in a real edge (sentiment as a short-term price catalyst in high-coverage names). The PSR above 0.9 is a non-trivial result given limited trades.
Risks: The strategy is essentially dormant across most market regimes. Three silent folds suggest the sentiment signal either isn't triggering, or is triggering and being filtered out. With only 2 closed trades in the backtest, no statistically reliable win rate can be computed. The DSR flags overfitting risk from trial count. Until the agent demonstrates consistent signal generation across varied market conditions, live-capital allocation should remain conservative.
What to Watch
The key question over the next 30–60 days is whether the agent begins firing in the current market environment. If daily runs continue producing zeros, the signal pipeline — news source freshness, sentiment model calibration, or entry thresholds — warrants inspection before any capital increase.