Give your AI coding assistant (Cursor, Claude Desktop, Windsurf, Antigravity) or custom LangChain/AutoGPT agents institutional-grade quantitative market intelligence.
Standard LLMs hallucinate financial prices and have zero understanding of market microstructure. This Model Context Protocol (MCP) server provides your AI agents with real-time deterministic quantitative tools over standard stdio JSON-RPC 2.0.
TOOLS INCLUDED IN THE MCP SERVER:
1. get_market_regime(symbol): Classifies asset microstructure into 6 discrete states (UPTREND, DOWNTREND, RANGE, VOLATILITY_EXPANSION, VOLATILITY_COMPRESSION, RISK_OFF) with calibrated probability simplexes.
2. get_byzantine_consensus(symbol): Fetches median real-time price consensus across Binance, Coinbase, and OKX with MAD outlier rejection.
3. calculate_liquidation_simplex(entry, leverage, direction): Computes liquidation threshold and survival probability simplexes.
4. portfolio_simplex: Multiclass Brier score calibration and fractional Kelly position sizing algorithms.
DELIVERABLES:
- Standalone Python package installable via pip install -e .
- Ready-to-use claude_desktop_config.json and .cursorrules integration snippets.
- 9 verified unit tests ensuring protocol compliance.
- Complete documentation with input/output schemas.
Instant delivery. Connect it to Cursor or Claude Desktop in 60 seconds.
