Anthropic Claude Powered • Machine-readable sentiment for active capital

Fast-flow market intelligence extracted from raw commentary.

VeloSignals continuously scans fragmented financial forums, social discourse, and macro publications. We extract verifiable conviction metrics, eliminate noise, and compile structured consensus time-series for tactical asset allocation.

Live Ingestion Stream

Macro & Single-Stock Sentiment Matrix

Latency: 180ms Model: Claude 3.5 Haiku + Sonnet

US10Y / Fixed Income Bullish Consensus (+0.68)

Consensus shifting toward term-premium compression. Growing agreement that long-duration treasury yields offer asymmetric risk-reward despite sticky headline CPI prints.

Semiconductors / AI CapEx Crowded Long / Divergence

Retail enthusiasm at peak percentiles, but institutional commentary expresses skepticism over tier-2 hyperscaler monetization and free-cash-flow ROI.

BTC / Liquid Macro Neutral Consolidation

Derivatives positioning neutral. Organic conviction low; market awaiting institutional flows confirmation and liquidity impulse.

SCHEMA EXTRACTOR (Normalized JSON Output) Status: 200 OK

{
  "asset_id": "US10Y",
  "timestamp": "2026-10-20T21:40:00Z",
  "sentiment_score": +0.68,
  "polarity": "BULLISH",
  "consensus_dispersion": 0.24,
  "institutional_divergence": false,
  "dominant_catalysts": [
    "yield curve normalization",
    "duration lock-in"
  ],
  "extraction_model": "claude-3-5-haiku-20241022",
  "confidence_interval": 0.93
}

01 / Ingestion

Continuous Noise Decoupling

Broad-spectrum scraping of retail threads, investor letters, and macro forums. Bot traffic and copy-paste engagement farming are aggressively filtered.

02 / Synthesis

LLM-Driven Consensus Mapping

Claude models extract underlying reasoning rather than just raw keywords. Sarcasm, contrarian theses, and changing stances are quantified systematically.

03 / Delivery

Queryable API & Time-Series

Export signals directly to SQLite/PostgreSQL, trigger automated alerts, or query the REST endpoints to feed algorithmic execution models.