Market + filing analysis
FINENGINE
A full-stack company-intelligence platform connecting live market telemetry, SEC fundamentals, peer benchmarking, filing evidence, and transparent valuation scenarios.
FINENGINE turns fragmented public-company research into one traceable workflow: market telemetry, SEC-normalized fundamentals, peer comparison, filing evidence, and transparent valuation scenarios. It is an analytical tool, not a stock-picking engine.
Company research is usually split across market-data sites, spreadsheets, valuation work, watchlists, and SEC filings. That makes even a straightforward company review more fragmented than it should be.
A React/TypeScript and FastAPI application that combines Twelve Data market telemetry with SEC EDGAR/XBRL fundamentals, peer comparison, filing evidence retrieval, watchlists, and assumption-driven valuation scenarios.
What the system contains.
FastAPI backend for upstream credentials, SEC retrieval, normalization, analytics, caching, filing extraction, and valuation math.
React/TypeScript interface for company overview, peer comparison, filing search, watchlists, and scenario analysis.
SEC XBRL normalization for revenue, earnings, cash, debt, equity, cash flow, capex, shares, and derived ratios.
Filing Lens for deterministic retrieval of evidence from recent 10-K and 10-Q documents.
Multi-stage Docker deployment that builds the Vite client and serves the SPA and API from one FastAPI origin.
System flow.
Technical decisions.
Use SEC EDGAR/XBRL as the source of truth for company-reported fundamentals while Twelve Data handles market-price telemetry.
Normalize annual flow metrics separately from point-in-time balance-sheet facts so valuation ratios do not silently mix incompatible periods.
Use deterministic filing retrieval to return evidence passages and direct SEC links before adding any optional LLM layer.
Expose DCF assumptions directly instead of hiding valuation behind an opaque score or recommendation.
Free and low-cost market-data sources vary in coverage, freshness, and rate limits. The application has to surface those boundaries rather than presenting every number as equally authoritative.