The Thesis
The news-sentiment strategy is straightforward by design: buy when recent news flow around a security turns positive, exit when sentiment flips negative. It runs across a 24-symbol large-cap universe spanning tech, financials, healthcare, consumer staples, industrials, and energy — a diversified spread intended to keep the signal from becoming a sector bet.
Simplicity is a feature here, not a limitation. Sentiment-driven strategies are easy to reason about, audit, and explain — qualities that matter in a paper-trading environment where understanding why a position was taken is as important as whether it worked.
Recent Activity
The strategy has been running daily scheduled sweeps, but the last week of logs (June 19–25) shows zero executions across every session. Cash sits at $2,810.08 against a total portfolio value that has drifted from $9,748 down to $9,606 — reflecting mark-to-market moves on existing positions rather than new decisions.
The most recent trade was a 4-share buy of UNH at $402.85 on June 18, following earlier buys of BAC (35 shares, $55.93, June 12), MSFT (4 shares, $428.23, June 3), and AAPL (6 shares, $312.06, May 31). All four positions were triggered by positive news reads, and none have been exited — suggesting sentiment has remained broadly constructive on these names, or that exit conditions haven't been triggered.
The silence over the past week is either a sign that the model is correctly finding no strong positive signals in a choppy tape, or that the signal threshold is set conservatively enough to suppress marginal calls. Both interpretations are reasonable; the data doesn't distinguish them.
Backtest and Validation Performance
Across a 451-day backtest window, the strategy produced a +0.19% total return with a 0.74 Sharpe ratio and a shallow 5% max drawdown. Fees were minimal ($2 total), and there were only 2 trades counted in the backtest summary — a notably low turnover for a sentiment-driven system operating over more than a year.
The cross-validation picture is more revealing. Across four folds spanning August 2024 through May 2026, only fold 4 (December 2025–May 2026) produced any activity or return. Folds 1 through 3 returned zero trades and zero return each. Fold 4 alone accounts for the entire +0.19% return and the out-of-sample Sharpe of 1.5 — a notably strong number, but derived from just two trades.
The formal validation gate did not pass. The Probabilistic Sharpe Ratio sits at 0.923 (below typical thresholds) and the Deflated Sharpe Ratio at 0.551 — the latter adjusting for the fact that six parameter trials were run, which inflates the risk of overfitting. With only 1 of 4 folds positive, the strategy hasn't demonstrated consistency across market regimes.
Strengths and Risks
Strengths: The drawdown profile is excellent — 5% maximum loss is very contained for an equity strategy. The most recent fold's 1.5 Sharpe is genuinely promising, and the large-cap, liquid universe avoids execution risk. The strategy is also interpretable: every trade has a news-driven rationale.
Risks: The sample is thin. Two backtest trades and near-total inactivity across three of four validation folds make it impossible to distinguish skill from luck. The failed DSR suggests the positive result in fold 4 may not survive a multiple-comparisons adjustment. The week-long quiet period could indicate a signal that only fires in specific market conditions — which would need to be understood before scaling.
Bottom Line
News-sentiment is a compelling concept with early signs of life, but its validation history is too sparse to draw strong conclusions. The priority for the next phase should be understanding why the signal was dormant for 18 months before activating — and whether that behavioral gap reflects a genuine market condition or a calibration issue. More trades, more folds, and a longer live window will be needed before this one earns a larger allocation.