News Sentiment scores the average tone of recent headlines about a ticker, from -1 (bearish) to +1 (bullish), over a trailing 7-day window. Each headline is read and scored by a language model; the score you see is the average across them.
Naive sentiment tools routinely misread financial language. "Missed estimates" and "beat expectations" both look like ordinary prose to a scorer trained on generic text, yet they carry opposite financial meaning. So does "guidance cut" versus "margin expansion", or "shares fall on record profit" — a headline that is simultaneously good news and a bad day.
The scoring model is therefore asked for two things per headline: a direction (bullish, bearish or neutral) and a conviction — how strongly it holds that view. The two are multiplied to produce that headline's score, so a hedged, ambiguous headline lands near zero rather than being forced into a side. The averaging then happens across those signed scores.
News sentiment reflects what has already been published, not what will be published next — a name with strongly negative 7-day sentiment can turn on a single subsequent headline, and sentiment from a thinly covered ticker is noisier than from a heavily covered one. Quantustik treats it as a supplementary signal alongside price action and fundamentals.
Neutral — recent headlines about the ticker skew neither bullish nor bearish on average over the trailing 7-day window.
A language model reads the headline and returns both a direction (bullish, bearish or neutral) and a conviction. The two are multiplied, so a hedged or ambiguous headline scores near zero instead of being forced onto one side.
No — it reflects already-published coverage, not future headlines, and can reverse quickly. It's a supplementary input, not a forecast on its own.
AAPL analysis shows this metric in context, or browse all S&P 500 tickers.
Educational research only — not investment advice.