Market Oracle delivers clean, verified market data to trading systems and AI agents: closed bars, indicators, order-book microstructure, macro context, and market regime. One request. You make the decisions — we supply facts only.
Oracle does not open trades or manage positions. Its job is to give your algorithms correct data you can build decisions on.
Trading decisions are formed exclusively on a closed 1m bar (closed: true). No look-ahead — data is always point-in-time correct.
EMA, RSI, ATR, VWAP, OBV, MFI, CVD, and ADX are covered by independent batch-reference tests. Closed bars and indicators are stored atomically; state recovers after failures.
One call to GET /v1/context/{symbol} returns everything for a decision: bar, order book, derivatives, macro, and market regime. Machine-readable formats, explicit limits, predictable errors.
The main endpoint for bots and AI agents. Bar with indicators, ticker, live order book, derivatives (funding/OI/liquidations), macro snapshot, venue consensus, market regime, and data_quality — instead of 8–10 separate calls.
GET /v1/context/{symbol}The only signal source is a closed 1m bar (closed: true). Everything else is context for sizing, filters, and pausing before macro events. Gate: ready, indicators_ready, data_quality.score ≥ 70.
Incremental computation without replaying history. Values are decimal strings; fields may be null during warmup.
Closed minute aggregation of top-20: spread, imbalance, walls, pressure.
/v1/microstructure/{symbol}Fear&Greed, dominance, stablecoins, DXY, 10Y yield.
/v1/macroFunding, open interest, liquidation cascades.
/v1/derivatives/{symbol}Correlation with BTC, realized vol, market breadth.
/v1/market-regimeMedian-mid across Binance, Bybit, and OKX with divergence control — protection against a bad tick before entry.
/v1/quotes/{symbol}bar_close + opt-in microstructure_close; REST to recover gaps.
WS /v1/streamUpcoming events (FOMC, CPI) with impact level.
/v1/calendar1m — 21 days with indicators, 1h — ~9 mo, 4h — 2 years, 1d — 3 years, 1w — 5 years.
/v1/history/{symbol}?interval=…A typical flow for an AI agent or trading bot.
Confirm status=ok, Redis is up, and the WebSocket provider is connected.
Daily limits, RPM, WS count, and key expiry — so you can pick a request cadence.
Take only ready=true, ideally history=full_day — history is warmed up.
Receive bar_close in real time. Order book optionally via microstructure:true.
Every 60–90 s per active pair — depth, macro, regime, and data quality in one response.
Exactly once per closed bar, with deduplication by symbol:ts.
Get an API key — and go. Everything is managed from one page.
Pause = steady pace under the RPM limit (60 000 ms ÷ RPM). Already have a key?
The manual is written so an AI agent or developer can connect Oracle without extra questions.
// Minimal Market Oracle client const res = await fetch( `${BASE}/v1/context/BTCUSDT`, { headers: { Authorization: `Bearer ${KEY}` } } ); const ctx = await res.json(); // Signal — only on a closed bar if (ctx.bar.closed && ctx.data_quality.score >= 70) { decide(ctx.bar, ctx.depth, ctx.market_regime); }
Data and strategies live apart: Oracle supplies facts; the Lab turns them into decisions.
Clean market data: bars, indicators, order book, macro, and market regime.
● You are hereStrategies, backtests, and a decision center on top of Oracle data.
Coming soon