Intelligence

Market scans and pair setups from Superior's analysis engine.

Two endpoints expose the analysis layer that powers the Superior Terminal's own market read. They return computed opinions with receipts — every claim carries the underlying numbers so an agent can weigh it rather than obey it.

GET/context/scan

Ranked market scan: which instruments currently look interesting, by category.

Query parameters

ParamTypeNotes
bucketstringUniverse to scan: alts, hip3, all — majors (BTC/ETH/SOL) are excluded by design: the scan hunts dislocations, and majors are the benchmark they're measured against
categorystringmomentum, mean_reversion, breakout, volume, all
top_nintDefault 10, max 50

Response — 200

json
{
  "computed_at": "2026-07-28T10:40:00Z",
  "results": [
    {
      "symbol": "SOL/USDC:USDC",
      "category": "breakout",
      "score": 0.81,
      "evidence": {
        "range_days": 12,
        "distance_to_range_high_pct": 0.4,
        "rvol_20": 1.9,
        "funding_annualized": 0.08
      }
    }
  ]
}

GET/context/setup/:pair

The current technical picture for one instrument: trend state, key levels, volatility, funding posture.

Response — 200

json
{
  "symbol": "BTC/USDC:USDC",
  "computed_at": "2026-07-28T10:40:00Z",
  "regime": { "state": "range", "adx_14": 17.2 },
  "levels": { "support": [66100, 64800], "resistance": [68950, 70200] },
  "volatility": { "atr_14_pct": 1.8, "bb_width_pctile_90d": 22 },
  "funding": { "current_annualized": 0.06, "trend": "flat" }
}

What this is, and isn't

Scores are deterministic computations over Superior's live market data — not model output, not advice, and not a signal service. The evidence object is the point: an agent should read the numbers, apply its own thesis, and backtest before capital moves. Responses refresh on engine cadence (minutes, not ticks); computed_at tells you exactly how fresh your copy is.