kavi-lin/market-sentiment-analyzer

Market Sentiment Analyzer

Purpose

Replaces "Web search: CNN Fear Greed Index today" used by older protocols. Returns a composite sentiment score (0-100) from multiple technical sentiment inputs, so the investment and sector protocols can consume one stable JSON structure.

Usage

python3 skills/market-sentiment-analyzer/scripts/sentiment.py                 # uses 15-min cache if fresh
python3 skills/market-sentiment-analyzer/scripts/sentiment.py --json-only
python3 skills/market-sentiment-analyzer/scripts/sentiment.py --no-cache      # force fresh compute
python3 skills/market-sentiment-analyzer/scripts/sentiment.py --max-age 300   # custom TTL seconds

No API key required. Uses yfinance (VIX, SPY, SPY puts/calls proxy) and requests to fetch CNN Fear & Greed from the public alternative.me endpoint (crypto F&G as proxy) AND attempts the CNN backend JSON endpoint.

Caching (default behavior)

Market sentiment is ticker-agnostic — when the investment protocol analyses several different tickers in one session, every Sentiment subagent would otherwise call this script fresh. Results are cached to skills/market-sentiment-analyzer/cache/sentiment_latest.json with a default TTL of 900 seconds (15 min). Within that window the full JSON (including cache_hit: true and cache_age_sec) is returned immediately — no network calls, no yfinance download. Override with --no-cache for truly-live values.

Output schema

{
  "generated_at": "ISO8601",
  "composite_score": "float 0-100 (0=extreme fear, 100=extreme greed)",
  "label": "Extreme Fear | Fear | Neutral | Greed | Extreme Greed",
  "vix": {
    "current": "float",
    "percentile_1y": "float 0-100",
    "regime": "LOW | NORMAL | ELEVATED | CRISIS"
  },
  "spy_momentum": {
    "rsi_14": "float",
    "pct_above_ma50": "float",
    "pct_above_ma200": "float"
  },
  "put_call_ratio": "float (null if fetch failed)",
  "fear_greed_index": "float 0-100 (null if fetch failed — fall back to composite)",
  "components_used": ["vix", "spy_rsi", "pct_above_ma", "pcr", "fg"],
  "extreme_sentiment_triggered": "true | false (composite > 80 or < 20)",
  "cache_hit":      "true | false (added by cache layer)",
  "cache_age_sec":  "int — seconds since cache was written (0 on fresh compute)"
}

Composite calculation

Equal-weight average of available normalized components. Each component maps to 0-100:

• VIX: inverted — lower VIX = higher greed. VIX 12 → 90, 20 → 50, 35 → 15
• SPY RSI(14): linear — RSI 30 → 30, 50 → 50, 70 → 70
• SPY vs MA200: 0% → 50, +10% → 80, -10% → 20
• Put/Call ratio: inverted — 1.2 → 20, 0.9 → 50, 0.6 → 85
• CNN F&G: direct passthrough

Missing components are dropped from the average (not imputed).

Extreme sentiment trigger

composite_score > 80 → Greed extreme → sector/investment protocols add extreme_sentiment risk flag composite_score < 20 → Fear extreme → contrarian long opportunities flagged