The thesis
news-sentiment runs one of the most legible rules on the platform: buy names showing 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), healthcare (JNJ, UNH, PFE, ABBV), staples (PG, KO, WMT, COST) and energy/industrials (XOM, CVX, CAT). The attraction is interpretability — every position maps to a stated news signal — and the discipline of a built-in exit. The open question is whether the signal fires often enough, and precisely enough, to matter.
Recent activity
Since the late-October restart the strategy has been a buyer only. It opened V, NVDA and PFE on 29 October, added KO (21 shares at $95.13) on 3 November, and AAPL (5 shares at $348.52) on 10 November. No exit has yet executed — the "sell on negative sentiment" leg remains untested in live trading. Most scheduled runs are quiet, logging 0 executed trades. The book has drifted lower through the month: total value fell from $9,987.62 on 4 November to $9,538.59 on 10 November, while cash was drawn down from $2,387.18 to $626.15 after the AAPL buy.
Backtest and validation
The backtest is where I urge caution. Over 451 days it returned just 0.19% (final equity $10,018.55, CAGR ~0.1%), with a Sharpe of 0.74, a trivial 0.05% max drawdown, 36.53% turnover and only 2 trades at $2 total fees. A strategy that places two trades in roughly eighteen months is barely expressing a view.
Validation did not pass. Across 4 walk-forward folds, only 1 was positive — and folds 1 through 3 placed zero trades, so the entire result rests on fold 4's two trades (which did post a tidy 1.5 Sharpe). The probabilistic Sharpe ratio looks reassuring at 0.923, but the deflated Sharpe ratio — which penalises for the 6 trials run — drops to 0.551. The honest reading: one lucky window, not a repeatable edge.
Strengths
- Capital preservation: drawdowns are tiny and the strategy sits in cash when it sees nothing.
- Low cost: negligible fees, no FX drag.
- Clarity: the thesis and the one promising out-of-sample fold (Sharpe 1.5) are easy to reason about.
Risks
- Sample size of two. No statistical confidence is possible here; win rate reads 0%.
- The signal rarely triggers, so most of the universe goes untraded and most runs are empty.
- The exit rule is unproven — no sell has fired live or in three of four folds.
- Failed robustness gate, with live equity currently below the backtest's final value.
Verdict
Keep it live but small. The logic is sound and cheap to run, yet the evidence base is far too thin to size up. The next useful data point is simple: enough trades — including a real exit — to tell signal from noise.