Deterministic comparable-company
analysis from real SEC data.
- Same multiples definitions across every peer.
- Same fiscal alignment, disclosed in every chain.
- Same refusal language when sectors don't match.
Bloomberg ships black-box multiples.
CIPHER ships YAML you can read.
~600ms debounce · then auto-resolves SIC-matched peers from EDGAR
Comps ask: how does the market value similar companies?
A comparable-company analysis lines up a target company against public peers and compares valuation multiples like EV/Revenue, EV/EBITDA, and P/E using the same definitions across the whole peer set.
What it is for
Bankers use comps as a market reality check. A DCF says what a company could be worth under assumptions; comps show what investors are currently paying for similar revenue, earnings, and cash-flow profiles.
Why it usually takes hours
A junior banker can spend 2-8 hours building a clean public comps table: choosing peers, pulling filings, aligning fiscal periods, fetching market data, calculating multiples, and explaining every excluded cell.
Why a normal LLM is risky
A chatbot may mix peer definitions, use stale market caps, mismatch fiscal years, or invent missing denominators. CIPHER uses sealed multiple definitions, SEC-backed facts, fiscal-alignment flags, and explicit refusals when a comp is unsafe.
Five verified pipeline stages. Same methodology as the DCF flow.
- RESOLVE · ticker → SIC → peer cohort (auto via SEC submissions; chips override)
- EDGAR fetch · bulk Company Facts pull for target + peers
- ALIGN · LFY method per fiscal_alignment.yaml; misaligned peers flagged
- MARKET · yfinance with explicit timestamp + URL provenance
- COMPUTE · 13 multiples × N peers; refusals carry corpus-verbatim language