Cross-asset sentiment index

Built entirely from real financial news headlines, scored by FinBERT — not trader positioning, order flow, or social-media chatter like most sentiment indices you'll find elsewhere. One number for crypto, FX, stocks, ETFs and commodities together, plus a breakdown by asset class below.

Why this isn't another Fear & Greed Index

Most sentiment indices — alternative.me's Crypto Fear & Greed Index, CFGI, the Daily Sentiment Index used by futures traders, broker-side indicators like Finlogix or FastBull — are built from trader positioning (long/short ratios), price momentum, or volatility. They measure what traders are doing.

This index measures what's being written. Every headline across 42 tracked tickers is scored by FinBERT, a transformer model fine-tuned specifically on financial text — the same model and pipeline behind SentimentFX's dashboard and API. It's a different input entirely, which is why the two kinds of index can (and often do) disagree.

How the score is calculated

Each ticker gets a 7-day trailing average of its FinBERT sentiment scores (-1 to +1, mapped onto a 0-100 scale where 50 is neutral). Tickers are grouped into five asset classes and equal-weighted within each class, so high-volume tickers like Bitcoin don't drown out lower-coverage ones like Dogecoin. The five class scores are then equal-weighted into the overall number — equities have 20 tracked tickers against commodities' 4, so a straight average across all 42 would just be an equity number wearing a cross-asset label.

API access

The same data is available as JSON at GET /sentiment-index — no authentication required, free to embed or cite. See developers.sentimentfx.org for the full per-ticker API (sentiment history, prices, correlation) and the MCP server.