Htx Sentiment Analyst

HTX Sentiment Analyst

Layer 2 analytical skill that reads market mood from 3 dimensions: market-wide fear/greed, smart-money positioning, and breadth (gainers vs losers). Calls one free external API (alternative.me) and composes 2 HTX Layer 1 skills.

When to use this skill

• "What's the market sentiment right now?"
• "Are we in greed territory?"
• "Is BTC positioning crowded?"
• "Is the crowd net long or net short?"
• "Should I be contrarian here?"
• "Sentiment vs price divergence?"

For pure derivatives crowdedness on a specific contract, prefer htx-derivatives-analyst. For pure technicals, prefer htx-technical-analysis.

Underlying tools

If those Layer 1 skills aren't installed:

htx-cli skill install elite-positioning
htx-cli skill install spot-market

Standard workflow

Step 1 — Fetch Fear & Greed (external)

curl -s 'https://api.alternative.me/fng/?limit=30' | jq .

Returns a 30-day history of values 0–100 with labels:

Step 2 — Pull HTX elite positioning (BTC as proxy for the major-coin crowd, plus user-specified symbol)

htx-cli futures call GET /linear-swap-api/v1/swap_elite_account_ratio \
  -p contract_code=BTC-USDT -p period=4hour --json

htx-cli futures call GET /linear-swap-api/v1/swap_elite_position_ratio \
  -p contract_code=BTC-USDT -p period=4hour --json

Step 3 — Pull breadth from spot tickers

htx-cli spot market tickers --json

Compute client-side:

• count of pairs with change > +5% (strong gainers)
• count with change < -5% (strong losers)
• gainer_to_loser_ratio
• top 5 gainers, top 5 losers

Composite sentiment score

Blend three dimensions to a 0-100 score (higher = greedier / more bullish crowd):

Weighted average → label using same Fear/Greed buckets.

Crowdedness interpretation

Cross-reference sentiment with price for divergence signals:

Output structure

{
  "skill": "sentiment-analyst",
  "timestamp": "2026-...",
  "summary": {
    "sentiment_score": 0-100,
    "sentiment_label": "Extreme Fear | Fear | Neutral | Greed | Extreme Greed",
    "crowd_direction": "long-leaning | balanced | short-leaning",
    "divergence": "sentiment-price diverging | aligned | none observed",
    "one_liner": "Greed (74), elite account ratio 1.6 long-leaning, 7d gainer/loser 1.8, but BTC price flat — distribution risk"
  },
  "fear_greed": {"value": 74, "label": "Greed", "change_24h": 5, "change_7d": -3},
  "elite_position": {
    "btc_account_ratio": 1.6,
    "btc_position_ratio": 1.4,
    "label": "moderately long-leaning"
  },
  "breadth": {
    "strong_gainers": 35,
    "strong_losers": 19,
    "gainer_loser_ratio": 1.84,
    "top_gainers": ["XYZ +12%", "..."],
    "top_losers": ["...", "..."]
  },
  "risk_warning": "Sentiment > 70 historically precedes 5%+ corrections within 7 days ~40% of the time."
}

What this skill explicitly does NOT do

• ⚠️ No social media sentiment (Twitter/X, Reddit, Weibo, Douyin) — HTX has no native endpoint and we're not paying for LunarCrush yet.
• ⚠️ No KOL monitoring — same reason.
• ⚠️ No on-chain whale flow — needs external paid source (Whale Alert / Nansen).

These gaps are flagged in the output risk_warning so users know what's not modeled.

Related skills

• https://github.com/htx-exchange/htx-skills-hub/elite-positioning, https://github.com/htx-exchange/htx-skills-hub/spot-market — data sources
• https://github.com/htx-exchange/htx-skills-hub/derivatives-analyst — combine with derivatives pressure for fuller picture
• https://github.com/htx-exchange/htx-skills-hub/market-overview — uses the same breadth computation in a different framing