Thesis
The news-sentiment strategy applies a simple, intuitive rule to a 24-stock universe of large-cap US equities: enter on strongly positive news sentiment, exit when sentiment turns negative. The universe spans six sectors — technology (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), and industrials (CAT, HON, DIS). The breadth is intentional: by watching household names across sectors, the agent can rotate into whichever corner of the market news flow currently favours.
Recent Activity
The strategy has been running scheduled daily sessions since at least late May. The journal tells a quiet story: two trades executed, no rejections, and otherwise a wall of zero-trade runs.
- 2026-05-31 — Bought 6 shares of AAPL at $312.06.
- 2026-06-03 — Bought 4 shares of MSFT at $428.23.
- 2026-06-04 through 2026-06-10 — Five consecutive sessions with no executions.
The portfolio currently holds approximately $3,360 in equities against $6,415 in cash, for a total paper value of roughly $9,774 — down slightly from a recent high of $9,996 on 4 June. Both positions remain open; no exits have been triggered, which is why the reported win rate is zero (no closed trades to score).
The inactivity is not a malfunction. A threshold-gated sentiment strategy should sit on its hands when the signal is ambiguous — and an extended quiet spell after an early-June entry suggests the agent is reading the market's noise floor accurately rather than churning.
Backtest & Validation Performance
Over the full 451-day backtest window the strategy returned +0.19% on a $10,000 paper account, finishing at $10,018.55. The full-period Sharpe ratio is 0.74, annualised CAGR a thin 0.10%, and the maximum drawdown a tidy 0.05%. Turnover was 36.5% annualised — high relative to the two trades logged, reflecting portfolio mark-to-market volatility rather than excessive churn.
The walk-forward breakdown is where the picture gets complicated:
| Fold | Period | Return | Sharpe | Trades |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | 0.00% | 0.00 | 0 |
| 2 | Jan – Jul 2025 | 0.00% | 0.00 | 0 |
| 3 | Jul – Dec 2025 | 0.00% | 0.00 | 0 |
| 4 | Dec 2025 – May 2026 | +0.19% | 1.50 | 2 |
Only one of four folds is positive — and only fold 4 fired any trades at all. The out-of-sample Sharpe of 1.50 is genuinely impressive, but it rests on two data points. The Probabilistic Sharpe Ratio (PSR) comes in at 0.923, meaning there is a 92% probability the true Sharpe exceeds zero — a respectable figure. The Deflated Sharpe Ratio (DSR), which penalises for multiple testing across six trials, falls to 0.551, below the conventional 0.65 gate. The validation framework correctly flags this as a fail.
Strengths
- Capital preservation: a max drawdown of 0.05% over 15 months is excellent; the strategy is not bleeding during quiet periods.
- Fold-4 signal quality: when the agent did act, the risk-adjusted return was strong.
- High PSR: the underlying Sharpe has a good probability of being real, not noise.
Risks
- Insufficient trade history: two executed trades cannot support robust statistical inference. The DSR failure is a direct consequence of this.
- Signal sparsity: three of four walk-forward folds produced zero trades, raising the question of whether the sentiment threshold is too conservative or the news feed coverage is uneven across the universe.
- Open positions only: with no closed trades, there is no empirical win/loss rate to evaluate edge direction.
Outlook
News-sentiment is a live but data-sparse strategy. The right interpretation is cautiously encouraging: it has not lost money, it fires selectively, and its one active window showed a credible Sharpe. What it lacks is enough trade history to satisfy a rigorous multiple-testing correction. The strategy should continue running and accumulating signal. Revisit validation once it crosses ten or more closed trades.