Every HoundCreek memo starts from the same premise: a stock's value is not a single number, it's a distribution. Our job is to model that distribution honestly, map today's price onto it, and explain what drives its shape.
The pipeline runs in four disciplined stages. First, the raw material — SEC filings, market data, peer comps, and macro context — is reconciled into one structured snapshot of the business: what it sells, what it earns, and what the numbers actually say. Same primary sources on every memo; no headlines, no improvisation. Second, the valuation machinery runs. A reverse DCF reads the growth and margin path today's price already implies — a diagnostic, not the headline. A forward free-cash-flow simulation then runs thousands of modeled paths into the HoundCreek Fair-Value Distribution, with the Fair Value at its center.
The computation in a delivered memo is deterministic. The AI proposes the assumption ranges the valuation engines sample — bounded by deterministic sector and archetype constraints — and drafts the memo's prose; it may also apply a bounded, evidence-gated adjustment to a modeled fair value where the filings support one, disclosed in each memo's valuation note. The AI shapes the inputs; the math computes the numbers. For structurally distinct issuers — banks, insurers, REITs, cyclicals, pre-revenue names — the model's inputs are calibrated to a sector- or earnings-specific anchor before the distribution runs, and the memo names the basis it used. The full mechanics, and the limits of the method, are on our methodology page and in our disclosures.
Third, the evidence is weighed: signals, risks, and catalysts are drawn from the filings and scored against their actual support, while the reasoning layer audits the classifications and assumptions behind the model. Fourth, the memo is assembled — the snapshot, the distribution, and the analytical view merge into one institutional-style document, structured to the same template every time — and automated quality gates review the result before it is ever delivered.
The output is a research document, not a directive. Our rating framework — Outperform, Neutral, Underperform, and, where the evidence does not support a directional call, Speculative or Not Rated — describes where the modeled fair value sits relative to the current price, and it is a function of the model, not a trade recommendation. Memos are impersonal research for informational and educational use, produced by the same standardized process for every reader. We publish the assumptions, the data, and the limitations alongside every figure, because a number without its context isn't research.
We'll use this space to explain how the pipeline works and, just as importantly, how it changes as we improve it.