Every valuation on this site follows the same discipline: work out what a business actually earns for its owners, build a range of outcomes from evidence we can point to, subtract what the business owes, and only then compare the result with the share price. This page explains each step, and how to read the numbers on a company page.
Every figure traces to the company's own audited annual and quarterly SEC filings (10-K and 10-Q reports). We read the risk factors, legal proceedings and subsequent-events notes closely, because pricing agreements, settlements and patent expiries often appear only there.
Where a filing and an earnings call disagree, the filing wins. Management guidance, "adjusted" earnings and analyst estimates never set a number in our model. They may be useful context, but they are not audited and they are not evidence.
The quantity we value is owner earnings: the cash a business produces that could be paid to its owners without harming the business. We start from operating cash flow and then:
Each company page has an Owner's Earnings tab showing this figure per share, year by year, with the arithmetic and the source filing behind each number.
Our starting point is the median of owner earnings over a period that represents the business as it exists today: a full cycle for a cyclical company, or the years since a major acquisition or restructuring, with one-off distortions excluded.
We never start from the last twelve months. A weak recent stretch is often exactly why a share price fell, and anchoring on it would simply repeat the market's mood. Recent quarters shape the near-term outlook; they move the baseline only when several quarters in a row confirm the change.
Every business is discounted at 6%, roughly the 30-year US Treasury yield plus a small premium. We never raise the rate for a riskier business. Doing so simply imports the market's own view of risk, and a method that always agrees with the market can never find a bargain.
Risk is not ignored. It shows up explicitly in the spread between our bear and bull cases, where each risk has to be named and its effect estimated.
We publish three values per share: bear, base and bull. They are roughly the 10th, 50th and 90th percentile of how we think the business could turn out, discounted back to today.
Each case is one continuous story built only from drivers we can name and point to evidence for: a signed contract, a patent expiry, a customer concentration disclosed in the filings, a plant under construction. The bear case is the worst outcome those named risks support, not the worst outcome imaginable. A general "market recovery" is never a driver in any case.
We forecast ten years explicitly. After that, value continues only as far as something specific keeps the earning power going (a durable brand, a contract, a regulated position). Otherwise it fades to what demonstrably persists, or to zero.
We credit growth only when it is:
Even then we use a rate at or below the long-run record and taper it as the room to grow runs out. Hoped-for market growth, market-share gains, new products, margin expansion, turnarounds and management targets are left out. If they happen, they are upside we did not pay for.
We value the cash the whole business produces before paying its lenders, then subtract net debt: borrowings, preferred stock and firm commitments, less surplus cash. We adjust for debt that will cost materially more or less than 6%, and carry refinancing risk in the bear case.
A heavily indebted company can be worth less than nothing to its shareholders, and when that is the case we say so. That is why some values on this site are negative.
For banks and insurers, where borrowing is the raw material of the business rather than a way of financing it, we value earnings after interest instead.
The valuation is finished before we look at the share price, so a falling or rising price can't quietly shift our judgement.
When we update an analysis, we start from our previous work but challenge every assumption, test the key drivers against specific facts in the latest filings, and explain what changed and why.
The margin-of-safety value is set below our base-case value to leave room for error in our estimates. Its size is driven by the downside, not the upside:
It is a judgement call, not a formula output. It is not a price target, a prediction of where the share price will go, or a recommendation to buy. Where price alone isn't enough, we also name a specific business condition that would need to be met. That condition appears on the company page under "Business condition attached to the margin of safety."
We don't update every company on a fixed schedule. Our time is limited, so we put it where it matters most: companies whose share price is close to our margin-of-safety value. Where we believe a company is grossly overvalued, we revisit less often, because a fresh look is unlikely to change the conclusion.
Each company page shows the date of its most recent analysis. Market prices are checked several times a day, not continuously.
The bar shows the bear value at the left end and the bull value at the right, with the base case marked between them. An orange line marks our margin-of-safety value; a purple line marks the latest market price we have checked.
A simple, average or high rating for how hard the business is to understand from its filings. More complex businesses leave more room for error, and usually get a wider margin of safety.
We don't publish valuations for:
Searching for one of these tickers shows a short "not currently covered" notice instead of an analysis.
A conservative method is built to avoid overpaying, and that has costs. It will often value fast-growing businesses below their market price, because growth that isn't yet secured gets no credit. It will miss some turnarounds. And every estimate is a judgement call that can be wrong: each analysis explains what it assumes and what would have to be true for it to be mistaken.
Every valuation on this site comes from a system we designed: the method described on this page, the rules each analysis has to follow, and the evidence standards every number is held to. We use various AI models to assist us in drafting the text of the analysis that you read on this site.
No analyst, human or AI, gets everything right every time, which is why we never treat a valuation as finished. We are constantly working to deepen our understanding of the businesses we cover. Each time we revisit a company, we find something worth adding, whether that's a detail in a filing, a risk that deserves more weight, or a strength that was underrated. We believe our thesis comes out stronger with every pass, and that cycle of revisiting and refining is a deliberate part of how we work.
Everything written on this site about a company, including the bear, base and bull values and the margin-of-safety figure, represents our best conclusions from the information the company discloses in its filings. These conclusions are not predictions or guarantees, and the company's own filings remain the authoritative source.
Corrections and feedback. If you see something that looks wrong, or have a perspective we may have missed, please email [email protected] with the ticker and the passage in question. We're always glad to take a look.