How TickerMath computes its numbers

Where the numbers come from

Company figures come from the SEC — the filings themselves, in the structured form companies are required to publish. No data resellers, no scraped websites, and no AI generating figures. They are pulled from those filings and checked before anything publishes.

Valuations are built from audited annual filings only. Every figure behind a fair value comes from the annual report a company files once a year and its auditor signs — never from quarterly results, analyst estimates, or any figure no auditor has stood behind — so fundamentals change when a company files and not before. The cost is one we accept: a business changing faster than its own filing calendar goes temporarily unvalued rather than mis-valued, and where there is no current audited annual filing on record there is no coverage.

There is one exception, and the page it affects always names it. When a company splits its stock after its last annual report, the share count is re-based to the latest filed diluted figure. The alternative is dividing a pre-split count into a post-split price, which is how a company comes to look as though it earns more in a year than the whole of it is worth.

Share prices are the one thing filings cannot give us. Those are entered and verified by hand, and every read states the date of the price it used. A price is never live and never moves during the day.

How fresh is this?

Fundamentals change when companies file, and not before. A company reports quarterly at most, and the figures behind a fair value — a full year of cash flow, capital spending, debt and share count — arrive once a year in the annual report. There is no version of this page that updates faster, because there is no faster truth to report.

We check every covered company for new filings each night — every night following a US filing day, which is every night anything can have been filed. When a company files, its read is rebuilt and republished within a day, and anyone watching it gets a note explaining what moved.

Prices are refreshed by hand, about once a week, and expire after 30 days. They are the one figure no filing carries, so they are entered and checked rather than piped in — and because the cadence is human, every read prints the date of the price it used. Printing that date is not enough on its own: a fair value, a margin of safety and a verdict are all computed against the price, so a stale one does not make a read slightly out of date, it makes the conclusion wrong while every figure on the page still looks current. So a price older than 30 days stops being a price. The company's read is withheld exactly as if we had never had one, and it returns on the next refresh.

Nothing here is live to the second, on purpose. The value of a business changes when the business changes — when it wins customers, pays down debt, or files a worse year — not when its ticker twitches. A fair value that moved every second would be pretending the arithmetic behind it moved too. If seconds matter to your decision, you are trading, and this is not a trading tool.

The value read

We estimate what a business is worth by asking what cash it can hand its owners, and what that cash is worth today. Money arriving in ten years is worth less than money arriving now, so future cash is discounted back.

Free cash flow is operating cash flow less capital expenditures, both exactly as filed. It is the cash left after a company has paid to keep itself going and to grow.

Those three parameters — 5% growth, 2.5% terminal growth, 9% discount rate — are the whole model. They are deliberately unexciting and identical for every company, so a read reflects the company's own cash flows rather than our optimism about it.

The range beside each fair value is the same model run pessimistically and optimistically: 3% growth discounted at 10% at one end, 7% growth discounted at 8% at the other.

Net debt

We value the business, then subtract what it owes net of what it holds — borrowings, commercial paper and finance leases, less cash and marketable securities. What is left belongs to shareholders, and that is what gets divided by the share count.

The reason is simple: a shareholder stands behind the lenders. If a company owes $139B, the owners do not get the first $139B of value — the lenders do. Operating leases are shown but never counted as debt; they are a contract for future use rather than money borrowed.

This is one of the main reasons a TickerMath number runs lower than a site that values the business and stops there. On a heavily indebted company the two answers are miles apart, and every read carrying real debt says so in its own words.

When a year lies

A single year is not a level. Companies pay legal settlements, deposit money with tax authorities, settle an acquisition — and the year that happens in tells you about the event, not about the business.

So before valuing anything we compare the latest year of free cash flow against the company's own typical level: the median of the last 5 years. If the latest year sits more than 30% away from that and the company has not simply been travelling that way all along, the read uses the typical level instead, and the page says so in plain words — including the size of the difference.

The rule is symmetric. One unusually great year is treated exactly like one unusually terrible one, because a one-off high makes an expensive company look reasonable and that is the error that costs a reader money.

The second half of the test matters as much as the first. A company that is growing fast puts every year far above its own trailing median — that is what growth is, not a distortion, and its latest year stands. The same applies to a business in long decline: we do not average a real decline away to make it look better.

Implied growth: what the price is betting on

Turn the model around. Instead of asking what a company is worth, ask what growth rate would make today's price exactly right. That is the implied growth, and it is often the most useful line on the page: it converts "is this expensive?" into "do I believe this company can grow that fast?", which is a question you can actually judge.

The solver searches up to 40% a year. Past that we say more than 40% rather than invent a figure — quoting the edge of a search as though it were the answer would understate exactly the thing the line exists to point at.

The lights and the verdict

Five lights, each a rule with a stated threshold, no judgement involved:

• Valuation. Green when fair value sits more than 15% above the price, red when it sits more than 15% below, amber in between.
• What the price assumes. Compares the implied growth against the company's own three-year record.
• Revenue. Growing or shrinking, latest year against the one before.
• Cash flow. Whether free cash flow is compounding or flat-to-down.
• Balance sheet. Green with net cash; red when net debt is more than 4× annual free cash flow; amber between.

The verdict at the top is computed from those lights. Nobody types it. That is why it is sometimes blunt about a company everyone likes.

The moat read

A second lens, over ten fiscal years, asking whether a business has a durable edge rather than whether it is cheap today. 157 of the 241 companies with a value read also have a moat read; the rest say on the page which figure their filings do not carry. The two lenses ask different questions of different lines, so either can answer while the other cannot — and neither borrows the other's verdict.

• Return on invested capital. Operating profit after tax, divided by average invested capital — equity plus borrowings and commercial paper, less cash. Held above 15% in at least 8 of the ten years is the mark of a real edge; above 12% is solid but not dominant.
• When invested capital is too near zero, we don't show a return at all. A company with almost no capital on its books produces a return figure that swings from hundreds to thousands of percent on tiny changes — arithmetic, not a real return. When invested capital falls below 3% of revenue in any of the ten years, we withhold this one check and say so, rather than print a number we don't believe or invent a bigger denominator to tame it. A fabricated denominator is a fabricated number. The other four checks are unaffected.
• The one-dollar test. For every dollar the company kept rather than paid out, how much did profit rise?
• Margin stability. A gross margin that holds through a decade is pricing power; one that drifts down is competition.
• Share count. Falling means owners are getting a bigger slice; rising means the opposite.

The income read

A third lens, asking one question: can you trust this dividend? 217 of the 241 companies have an income read, 169 of which actually pay one — a company that pays nothing gets a plain card saying so rather than five checks about nothing.

• Coverage. The dividend against free cash flow, not against earnings — a payout funded by borrowing is a countdown whatever the earnings ratio says. Green at or under 60% of free cash flow, amber to 90%, red above it. It uses the same free-cash-flow base the value read discounts, so a one-off year cannot misjudge the payout either.
• The raise record. Raised, held or cut, year by year across the decade. Any cut is red — income investors remember cuts. Judged on the dividend a company declares per share rather than the cash that left in a given year, because a cut is a decision and cash timing is not: a payment landing the other side of a year-end is not a cut, and reading it as one would say a company did something it did not.
• Is the raise real. Ten-year growth in the dividend per share, stated plainly — green at 5%/yr or better. We do not tell you what inflation will be; that number is yours to judge against.
• Balance sheet. The same light the other lenses show. Debt is what kills dividends.
• Yield. Stated as fact, never celebrated. A yield more than 2× the typical 1.5% here, on a payout the cash does not comfortably cover, reads red: a yield gets high because the price fell, and the price usually fell for a reason.

Special dividends. A one-off payout would make the following year look like a savage cut. Where a year's dividends are at least half again both the year before and the year after, it contains a special, and the raise record steps over it and says so on the page.

The growth read

A fourth lens, on 225 of the 241 companies, asking how fast a business has grown and at whose expense. The only forward-looking number on that page is the one already inside today's price — what the market is charging for growth. We forecast nothing.

• Pace. Revenue per share over the decade and over three years, green at 8%/yr or better. Per share, because growth financed by issuing shares is not growth for someone who already owns them — the headline total sits beside it so the gap is visible.
• Consistency. How many of the nine year-steps rose, green at 7. Growth that shows up most years is a business; growth that shows up twice is an event, and a decade average hides the difference.
• Cash follows. Free cash flow per share against revenue per share. Growth that never reaches cash was bought rather than built.
• Returns survive it. Whether return on invested capital held up while the company grew. Where a filer's statements do not carry it, this one check declines and the other four still answer.
• The price's bet. The growth rate today's price implies, set against what the record delivered. Past 40%/yr the read says "more than this model will solve for" rather than quoting a number nobody has delivered.

A shrinking company gets real checks, honestly red, rather than no read at all — "this is not a growth company" is exactly what somebody asking about growth needs to hear.

What we refuse to do

241 companies have a read. 197 have a page that explains why they do not, and that is a feature we are proud of rather than a gap.

Banks, insurers, property trusts and regulated utilities get no fair value, because the free-cash-flow model does not describe them. A bank's cash flow is its business; a utility is supposed to outspend its cash flow. Running the model anyway would produce a confident-looking number that means nothing, and a confident wrong number is the one thing this site exists not to produce.

Some companies are held back for a duller reason: a figure we need is missing from the filings. Those pages name the missing number. We never estimate one to fill a gap.

Why our numbers often run lower than other sites

This is the question we get most, so here it is flatly. Three differences, all deliberate:

• We grow the filed record, not forecasts. Our growth rate is a fixed 5% applied to what a company has actually reported. Sites built on analyst estimates start from what a company is expected to earn, which is usually more.
• We discount free cash flow, not earnings. Earnings are an accounting result; free cash flow is what is left after the company pays for its own plant and equipment. For a capital-hungry business the gap is enormous.
• We subtract net debt. Many published fair values are the value of the business, full stop. Ours is the value of the business less what it owes.

None of that makes us right and them wrong. It makes the numbers different by design, with the assumptions on the table so you can decide which set you trust.

When we're wrong

We compute one model, honestly, and it can be wrong. Sometimes the model does not suit a company as well as we thought; sometimes we find a mistake in how a figure was read. When that happens we fix it and the numbers change.

That is the tool working, not the tool failing. It is also why watchlist alerts exist: when a number moves, the people following that company hear about it rather than discovering it by accident.

The disclaimer, in full

TickerMath is an educational tool. It is not investment advice, and it never tells anyone what to buy or sell.

• Everything here is for educational purposes only.
• Nothing on this site is investment advice, or a recommendation to buy or sell any security.
• TickerMath computes one model from public SEC filings — one model is a point of view, not a verdict.
• We make no guarantee that any figure is accurate or complete.
• Figures can contain errors, and they change as new filings and prices arrive.
• A fair value is an estimate produced by assumptions we publish; change the assumptions and the number changes.
• Past cash flows are not a promise about future ones.
• Do your own research, and consider talking to a licensed professional before you invest.
• TickerMath has no relationship with any company covered here, and no position to report in any of them.
• If you spot something wrong, tell us — we would rather fix it than defend it.

This is the same text as the disclaimer page.