Pick ticker and click Analyze.
Quick try:
Recency-weighted sentiment
—
Run to see verdict
—
−1.0 Very bearish0 Neutral+1.0 Very bullish
Sentiment distribution
📈 Sentiment Over Time
Each headline's score plotted by publish date. Trend line is the running 5-headline average. Recent headlines have more weight in the verdict.
📋 Headlines With Per-Headline Scores
Each headline scored separately. Click the title to read the full article. Pos words (green) and neg words (red) show exactly which terms drove each score — auditable, deterministic, lexicon-based.
🧪 Methodology
—
Lexicon-based ≠ neural sentiment. This engine counts positive vs negative words from a curated finance lexicon (~140 positive, ~210 negative terms, Loughran-McDonald-inspired). It handles negation ("not bad" → flips to positive) and intensifiers ("very strong" → 1.5x weight) but won't catch sarcasm or complex sentence structure. For long-form text (10-Ks, transcripts, analyst reports), pair this with the
/api/gemini AI endpoint for richer scoring.