Methodology
How StockValuer values a company
From raw SEC filings to three DCF cases: what we take as fact, what we research, what we compute, and which AI model does the reasoning.
Deterministic facts come from filings
The starting point is not an opinion. StockValuer pulls a company's 10-Q and 10-K financials directly from SEC EDGAR and keeps them current as new filings land. Base inputs — trailing revenue, diluted share count, net cash — are read from those filings, traceable to the source period. They are facts, not estimates.
Judgment inputs are researched, with citations
The forward-looking inputs a DCF needs — revenue growth, target operating margin, tax rate, reinvestment, and terminal growth — cannot be read off a filing. An AI model researches each one from primary sources and attaches a rationale, a confidence level, and citations to every figure.
This is the heart of the transparency promise: you see how each non-deterministic assumption was derived, for each of the three cases, rather than a bare number.
The discount rate is computed, not guessed
Rather than picking a round number, StockValuer computes the discount rate as a weighted average cost of capital (WACC). The cost of equity comes from CAPM — the risk-free rate, the company's beta, and the equity risk premium — and is blended with the after-tax cost of debt. Each of those market inputs is researched and shown, so the rate is auditable end to end.
Guardrails bound every published run
Researched values pass through a deterministic guardrail layer before they reach the DCF. If a figure falls outside a defensible range — a terminal growth rate above long-run GDP, say — it is clamped, and the adjustment is shown on the assumption. The goal is to keep any single run from producing an indefensible number.
Three cases, side by side
The pipeline produces a bear, base, and bull case, each with its own assumptions, computed discount rate, and intrinsic value per share. Comparing them against the market price yields the margin of safety — how far the stock trades below the computed value.
Which AI model reasoned it — and why that matters
Every valuation records the specific AI model that did the research, and StockValuer shows it — on the valuation, on the screener, and here. Frontier models bring broad, consistent, and citable reasoning to the judgment inputs, and naming the model lets you weigh that reasoning for yourself. Transparency about the reasoner is part of the product, not a footnote.
None of this is investment advice. It is a transparent, auditable estimate you can inspect, challenge, and — once signed in — edit into your own thesis.