About & Methodology
FilingSight publishes periodic, data-driven analysis of US-listed companies using SEC filings, Federal Reserve data, USDA commodity data and market prices.
How reports are built
Each company report is assembled by a pipeline of specialized analyst modules that consume public regulatory and market data. Every analyst emits a bullish/neutral/bearish signal, a confidence score and a structured justification. A composer weights them into a single composite rating.
Agentic layer (TradingAgents-inspired)
On top of the deterministic analysts runs a multi-agent graph: a Bull Researcher and Bear Researcher debate the case, a Trader forms a decision (action, conviction, price target), a Risk Manager enforces portfolio constraints and may veto, and a Portfolio Manager produces a personalized position size for your profile.
The graph works deterministically out of the box. When an LLM key (LLM_API_KEY) is configured, the trader's thesis and portfolio notes are enriched with richer reasoning — no other behavior changes.
analysts → bull/bear debate → trader → risk manager → portfolio manager
Data sources
Company facts (XBRL), filings & submissions, full-text search (13F/13D/13G).
GDP, CPI/PCE inflation, rates, unemployment, yield curve, VIX, Fed balance sheet.
Commodity production & prices (corn, soybeans, wheat, cattle).
Equity price history, 52-week range, market price (unofficial; best-effort).
SEC filing glossary
This report is generated automatically from public regulatory filings (SEC EDGAR), Federal Reserve data (FRED), USDA commodity data and market prices. It is informational and educational only, does not constitute investment advice, and has no regard for your personal circumstances. Verify all figures against primary sources. Investing involves risk of loss.