The thesis
news-sentiment runs a legible rule: buy names carrying positive recent news sentiment, and exit when sentiment turns negative. It trades a 24-stock large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), plus healthcare, staples and energy. The appeal is interpretability — every position should map to a readable catalyst. The flip side is that the strategy is only ever as good as the sentiment signal feeding it, and it only acts when that signal fires.
Live activity
The strategy is live, but lately it has been quiet by force rather than by choice. Across six scheduled sessions from 21 to 28 October, each run proposed exactly one trade and each was rejected — zero executions. Cash has sat unchanged at $2,036.71 while total equity drifted between roughly $10,321 and $10,453. The last actual fills were on 23 September (buy 4 NVDA at $225.34, sell 35 BAC at $55.96), following a cluster of May–July buys in AAPL, MSFT, GOOGL, UNH and BAC. A full week of rejected orders is worth flagging: the engine wants to act, but something — plausibly sizing or available cash — keeps blocking it.
Backtest and validation
Here the picture gets uncomfortable. Over a 451-day backtest the strategy returned just 0.19% (final equity $10,018.55, ~0.1% CAGR) on only two trades, with a Sharpe of 0.74 and a negligible 0.05% max drawdown. Fees were trivial ($2). Two trades over fifteen months is almost no evidence — the Sharpe, and the win rate reported at 0%, rest on a sample far too small to trust.
Validation did not pass. Of four walk-forward folds, three executed zero trades and returned nothing; the entire track record lives in fold 4 (Dec 2025–May 2026): +0.19%, Sharpe 1.5, two trades. One of four folds is positive, and that fold is the strategy's history. The deflated Sharpe (0.551) and probabilistic Sharpe (0.923) look superficially acceptable, but against six trials and a two-trade sample they carry little weight.
Verdict
The strengths are genuine but modest: a clear thesis, minimal drawdown, and one positive out-of-sample fold. The risks dominate. The signal fires so rarely that the backtest cannot separate skill from luck, and live execution is currently jammed, with a week of consecutive rejections. Until the rejection issue is diagnosed and the strategy produces enough trades to validate across folds rather than within one, news-sentiment should be read as an unproven candidate — an idea with a tidy rationale and not yet a demonstrable edge.