Market Context Overview
The data engine — what it serves and where the data comes from.
/context is the first pillar: everything an agent needs to understand a market before acting on it. It's read-only, fast, and backed by the same datasets Superior's own backtesting engine consumes.
| Endpoint | What it answers |
|---|---|
GET /context/venues | Where can I trade, with what framework, needing what? |
GET /context/markets | What instruments exist, and are they backtest-ready? |
GET /context/candles | How has this market moved? |
GET /context/funding | What does holding this perp cost? |
GET /context/datasets | Exactly what history exists, per venue and pair? |
GET /context/scan | What looks interesting right now? |
GET /context/setup/:pair | What's the current technical picture for this pair? |
GET /context/leaderboard | Returns the apps/api-owned paper-trading strategy leaderboard |
GET /context/traders | Planned; currently returns 501 feature_not_available |
Where the data comes from#
Historical data is served from immutable, validated dataset releases. Superior's backfill pipelines pull from each venue's own API, validate the result (gaps recorded, never silently bridged), freeze it into a checksummed release, and publish a pointer. /context/candles and /context/funding read those releases — which means:
- Reproducibility. A release ID pins the exact bytes. A backtest and your own analysis read identical history.
- Honesty. If a venue's own history has a hole (they do), the gap is reported in
/context/datasets, not papered over. - Cacheability. Immutable data means aggressive caching; repeated candle reads are cheap and fast.
Available intelligence signals (scan, setup) come from Superior's intelligence engine and refresh on engine cadence. Leaderboard and trader discovery remain unavailable until a production provider is implemented.
What context is not#
It is not a real-time tick feed. For live order-book and tick data, connect your deployed framework directly to the venue — that's what the runtime is for. Context answers research questions; the runtime handles execution-time data through the framework's own venue adapters.