Thesis
The news-sentiment strategy runs a simple, event-driven rule: buy names showing positive recent news sentiment and exit when sentiment turns negative. It operates over a 24-stock US large-cap universe — AAPL, MSFT, GOOGL, NVDA, JPM, BAC, V, MA, UNH, XOM, CAT and peers — so the thesis is less about stock-picking than about reacting quickly to the flow of headlines on quality names.
Recent activity
Lately the strategy has mostly sat on its hands. Across scheduled runs from 2026-10-13 to 2026-10-20, every run executed zero trades, with the two most recent runs each logging a single rejected order. Cash is pinned at $2,036.71 and the book's total value has drifted in a narrow $10,345–$10,510 band, last marked at $10,450.45. Its last real activity was in September: a 4-share NVDA buy at $225.34 and a 35-share BAC exit at $55.96 on 2026-09-23, following a cluster of summer entries in MSFT, GOOGL, UNH and AAPL. The signal, clearly, fires rarely.
Backtest & validation
The 451-day backtest is sobering. Total return was 0.19% (CAGR 0.1%), Sharpe 0.74, on just 2 trades, with a reported 0% win rate and $2 in fees. Max drawdown was a trivial 0.05% — but so was the return. In practice the strategy was flat.
Cross-validation makes the fragility explicit: it did not pass the gate. Of four walk-forward folds, three produced zero trades and zero return; all of the measured performance — 0.19% and a 1.5 out-of-sample Sharpe — comes from a single fold (2025-12 to 2026-05). A probabilistic Sharpe ratio of 0.923 looks reassuring, but the deflated Sharpe, which penalises the 6 trials run, sits at just 0.551 — barely above a coin flip. One positive fold out of four is not evidence of edge.
Verdict
Strengths: the strategy is cheap to run, disciplined, and risk-contained. It holds cash rather than forcing trades, and its minuscule drawdown means it has never meaningfully hurt the book.
Risks: these dominate. With only two lifetime trades there is no statistically meaningful record; the win rate is zero; and the validation gate rejected it precisely because the apparent returns rest on one lucky window. The recent rejected orders also hint at execution friction worth a closer look.
Bottom line: news-sentiment is live but effectively dormant, and the data we have cannot distinguish skill from noise. It is worth keeping as a low-cost, low-risk option, but it has not yet earned capital conviction.