Agent Financial Research Workflow

The Problem

You ask your coding agent to "research NVDA" and it Googles, finds three blog posts with outdated numbers, and gives you a confident summary with a P/E ratio that doesn't exist in any filing. You know what free cash flow, IV rank, and the 2s10s spread mean — you just need your agent to actually fetch and compute them from structured sources instead of guessing. Paid terminals (Bloomberg, FactSet) aren't an option for a solo developer who trades on the side. This skill teaches your agent to pull real data from yfinance, SEC EDGAR, and FRED, compute the metrics that matter, and synthesize a research packet you'd normally spend an hour assembling manually.

What You Get

• Fundamental data extraction via yfinance — P/E, P/B, EV/EBITDA, revenue growth, free cash flow, debt-to-equity, and 52-week range, with decision logic for pre-revenue or non-equity securities.
• SEC filing retrieval via the free EDGAR API — fetches 10-K, 10-Q, 8-K, and DEF 14A filings, extracts key sections, and flags material events without a paid subscription.
• Options market sentiment — computes put/call ratios, average IV for calls vs puts, and compares against 30-day historical averages to surface meaningful divergence.
• Macro context gathering via FRED — pulls treasury yields (2Y, 10Y), breakeven inflation, unemployment, and fed funds rate, with decision logic for growth vs value positioning.
• Structured research packet — synthesizes all data into a markdown report with sections for fundamentals (with interpretation), options sentiment, SEC filings, and macro context, plus a conditional recommendation framework.
• Multi-ticker comparison — run the same pipeline on two tickers and output a side-by-side comparison table.

Why This Beats Prompting It Yourself

Standard LLMs hallucinate stock prices and can't do real calculations. Ask for P/E and you might get a number from two years ago, or one that mixes up trailing vs forward. This skill forces yfinance + pandas for numbers, SEC EDGAR for filings, and FRED for macro — you get actual data with source attribution, not plausible guesses. The decision logic (if fundamentals strong AND sentiment bullish → favorable setup) is deterministic, not vibes-based.

Example Session

You: "Do a deep dive on TSLA — fundamentals, options sentiment, and macro context."

**Agent:

TSLA Research Packet: P/E 42.3x (sector median 25.1x), Revenue Growth +18.2% YoY, FCF $3.1B, Debt/Equity 0.12. Options: P/C ratio 0.82 (vs 30-day avg 0.91, mildly bullish), IV Skew elevated. SEC: latest 10-Q revenue $25.4B (+18%), 8-K new factory in Mexico $5B capex. Macro: 10Y 4.23%, Fed Funds 4.50%, curve normalized +34bp. Assessment: Strong fundamentals but premium-priced, macro neutral, CAUTIOUS entry.

Use Cases

• Pre-trade research for individual stock positions before sizing an order.
• Earnings season prep — pull the latest 10-Q and compute key metrics across your portfolio in minutes.
• Building a personal watchlist dashboard with automated fundamental + sentiment data.
• Comparing two stocks side-by-side for a sector rotation decision.
• CFA exam study — run the skill on real tickers to practice financial analysis frameworks.

Known Limitations

Requires yfinance for market data — won't work offline or in air-gapped environments. SEC EDGAR API requires a User-Agent header but no API key. FRED API requires a free API key from research.stlouisfed.org. Options analysis limited to US exchanges. The skill reads current market data at request time; it does not store historical data or maintain a database.