Octagon

Streams live market data, company fundamentals, and investment metrics through the Model Context Protocol (MCP) interface, enabling LLM agents and applications to access current financial information without polling. Implements MCP resource handlers that expose financial datasets as queryable endpoints, allowing Claude and other MCP-compatible clients to request specific securities, sectors, or market conditions with structured JSON responses.

Aggregates and normalizes private market data (venture capital, private equity, M&A) from multiple sources into a unified schema, exposing it through MCP endpoints. Implements data transformation pipelines that reconcile different data formats, handle missing fields, and standardize company identifiers across private market databases, enabling consistent querying across fragmented data sources.

Provides structured access to public market data including stock prices, financial statements, earnings reports, and valuation metrics through MCP tool and resource endpoints. Queries underlying financial data APIs (likely SEC EDGAR, Bloomberg, or similar) and returns normalized JSON responses with standardized field names, enabling LLM agents to retrieve company fundamentals without parsing HTML or handling API authentication.

Aggregates sector-level and broad market index data (S&P 500, Nasdaq, industry indices) through MCP endpoints, enabling queries for sector performance, composition, and comparative analysis. Implements index calculation and weighting logic, returning normalized sector metrics and constituent information that allows LLM agents to understand market structure and relative performance without manual index construction.

Leverages LLM reasoning capabilities through MCP to synthesize investment theses by combining real-time market data, fundamentals, and private market information into structured research narratives. The MCP server provides data access primitives that Claude or other LLMs use to build multi-step reasoning chains, generating investment recommendations with cited data sources and risk assessments without requiring pre-built templates.

Provides MCP tools for analyzing portfolio composition, calculating performance metrics, and attributing returns to specific holdings or factors. Implements portfolio weighting calculations, return aggregation, and risk metrics (volatility, Sharpe ratio, drawdown) by querying underlying security data and combining it with portfolio position data, enabling LLM agents to perform portfolio analysis without requiring external portfolio management systems.

Indexes and enables semantic search over earnings call transcripts through MCP, allowing LLM agents to retrieve relevant excerpts and perform textual analysis without downloading or parsing raw transcript files. Implements transcript storage with embeddings-based search, returning matched segments with speaker attribution and timestamp context, enabling agents to extract management guidance, Q&A insights, and sentiment signals from earnings calls.

Aggregates financial news, social media sentiment, and analyst commentary for securities through MCP endpoints, providing LLM agents with access to recent news, sentiment scores, and market commentary without requiring separate news API integrations. Implements news source aggregation and sentiment scoring (likely using pre-trained models), returning normalized news items with sentiment labels and source credibility indicators.

Enables peer group analysis by retrieving comparable company metrics and valuation multiples through MCP, allowing LLM agents to construct peer groups, calculate median multiples, and perform relative valuation analysis without manual data compilation. Implements peer selection logic (by sector, size, growth profile) and multiple calculations (EV/EBITDA, P/E, Price/Sales), returning normalized multiples with outlier detection and context for interpretation.