heurist-finance: Skill for Claude Code

Valuation Expectations

Shared plumbing and output contract: ../_shared/analyst-core.md. This skill runs valuation backwards — from price to implied expectations — because that's the only direction that produces a trade.

The core move

A DCF that outputs "fair value $147" is an opinion wearing a spreadsheet. The expectations-investing move (Rappaport & Mauboussin) inverts it: hold the discount rate and terminal assumptions fixed, and solve for what revenue growth / operating margin path makes today's price fair. Then ask one question: is that path plausible against base rates? The trade exists only where implied expectations and plausible outcomes diverge.

Workflow:

1. Pull the inputs: yahoo equity_overview symbol=SYM (price, market cap, forward estimates, analyst consensus), sec xbrl_fact_trends for the actual revenue/margin/FCF history (8–12 quarters + 5 annual), share count and net debt from the same. Cross-check any scraped estimate against equity_overview — scraped consensus numbers are frequently stale.
2. Solve for implied expectations with a small Python script (write it, don't estimate): a simple 10-yr FCF model at an 8–10% discount rate, terminal growth 2–3%. Anchor on TTM FCF built from the XBRL pulls — trailing-4q CFO minus trailing-4q capex (difference fiscal-YTD values where needed) — never a scraped "FCF" field on faith. Match PV to enterprise value (market cap + net debt) when the FCF is pre-financing firm FCF; match market cap only when modeling equity FCF — say which you used. The canonical solve is two-stage: take consensus estimates for years 1–2 as given (equity_overview) and solve for the growth the years 3–10 tail must deliver; quote the flat-CAGR solve alongside as the simple read (flat understates what's priced when near-term growth is front-loaded). Report the sentence: "At $X, the market is paying for ~Y% revenue CAGR at Z% operating margins for a decade."
3. Test against base rates. The reference-class question beats the inside view. Anchors from the base-rate literature (Mauboussin's corporate base-rate work): sustained 20%+ revenue CAGR over 10 years is achieved by roughly the top ~1–2% of companies; double-digit CAGR for a decade is roughly top decile, and the odds fall hard with starting revenue size ($10B+ companies almost never compound revenue >20% for 10 years). Margin expansion assumptions should be tested against the best margin the company — or its best-in-class peer — has ever printed.
4. Locate the variant view. State explicitly which implied assumption you disagree with, in which direction, and why you know something the consensus doesn't (structural change, misread cycle, forced sellers). No variant view → no trade, whatever the multiple says.