We stopped telling you what a stock is worth. Here is what the valuation tool does instead

Almost every valuation tool ends the same way: a single number, printed large, labelled fair value. Below it, a percentage gap against the market price. The reader is meant to do the obvious thing with that gap.

The problem is what sits underneath. That number is the output of a growth rate somebody picked, a discount rate somebody picked, and a terminal rate somebody picked. Change the growth assumption by three points and the fair value moves by a third. So the verdict was never in the math — it was in the assumptions, chosen by a stranger, and then hidden behind a dollar sign. A fixed forward assumption is a recommendation wearing a calculation as a costume.

We had that problem too. As of 2026-08-31 the valuation pages work the other way round.

What "the price implies" means

The market already contains a full set of assumptions. It has to — a price is what someone paid, and to pay it you had to believe something about future earnings. Those beliefs are recoverable. Take any valuation method, fix everything the filings give you, and solve backwards for the one input that would make that method print exactly today's price. Now the assumption is explicit, and you can judge it.

That is what every method on /valuation/AAPL now does by default. Here is Apple on 2026-09-02, at $325.13, against the FY2025 10-K (accession 0000320193-25-000079) plus the 10-Qs through Q3 FY2026:

Read that table as one sentence and it becomes a real question. At today's price, a discounted-cash-flow model only balances if Apple's free cash flow shrinks about 2.4% a year through stage one — or, holding growth where the filings put it, if you are content with a 7.2% required return. The Lynch method balances at 13.0% EPS growth for a decade. The filed EPS CAGRs on /growthrate/AAPL are 22.6% over one year, 6.8% over three, 17.7% over five and 12.4% over ten. So 13.0% is not an outlandish ask — it is roughly the ten-year record repeated for another decade. That is a claim you can have an opinion about. "Fair value: $163" is not.

The hero states it plainly and stops: Apple trades at $325.13. That price implies a P/E of 37.3×, 13.0% yearly EPS growth for a decade, 33.5× EV/EBITDA and a 2.8% free-cash-flow yield. Are those assumptions reasonable? ... the range you build is yours.

How to use it, in five steps

1. Open the symbol. /valuation/ plus the ticker. The page opens on the What the price implies strip — four tiles, each with its formula underneath (price ÷ trailing EPS, (market cap + net debt) ÷ EBITDA, and so on). Nothing is asserted about value yet.

2. Read the hero, then the filing behind it. Under the tabs sits The filing behind these numbers: revenue, diluted EPS, free cash flow, share count, the form, the period end, the accession number, and a link straight to sec.gov. Every input is traceable before you touch a slider.

3. Pick a tab. Seven of them: DCF, P/E growth (Lynch), Earnings power, Graham number, EV / EBITDA, EV / FCF, P/E vs sector. Each one opens with its sliders already sitting on the price-implied value. You are starting from the market's assumption, not from ours.

4. Drag to your own view. The growth sliders carry the company's own filed CAGRs as presets — for Apple's DCF tab: FCF CAGR 1-year −5.7%, 3-year −3.0%, 5-year 6.7%, 10-year 3.2%, plus an EvidInvest reference default. Pick a filed rate, or type your own. The reset link reads Reset to the implied number so you can always get back to the market's starting point.

Move a slider and a neutral delta line appears under the result: Your target P/E (25.4×) is below the 37.3× the price implies — your value lands 32% below the price. That is the whole verdict vocabulary. Direction and magnitude, no adjective. We removed the judgement words on purpose: the tool does not know your holding period, your required return or your alternatives, so it has no business scoring the outcome for you.

5. Save it or share it. Every input lives in the URL (?m=pe&g=eps5y&fg=17.7), so a link reproduces the exact scenario on someone else's screen. Signed in, Save stores the snapshot with a name and it reappears in the history panel. The clean URL always means implied; ?anchor=preset gives you the old behaviour, Range with preset inputs, which on Apple today lands the seven methods at $117–$202 against a $325.13 price.

The caveats, shown rather than hidden

Reverse-solving is honest, which means it sometimes produces a number whose job is to tell you the method does not apply.

Graham. Apple's implied Graham multiplier is 2,412 against Graham's classic 22.5. That is not a signal — it is book value per share of $5.03 doing what book value does to an asset-light company that has spent a decade returning capital. The Graham tab says so in its own description. Same story for the P/E-vs-sector tab: a 25.4× sector median applied to a company the market prices at 37.3× tells you about the sector's composition, not about Apple.

Financial and real-estate companies. Ask for JPMorgan's implied assumptions and three methods decline outright, each with its reason printed: Earnings power value is not used for financial or real-estate companies, and for both enterprise-value tabs, Enterprise-value multiples are not used for financial companies. What you do get is a P/E of 15.3× and an implied 9.8% EPS growth. Fewer methods, no fabricated ones.

Loss-makers and negative cash flow. A method that needs positive earnings or positive free cash flow cannot be solved back from a price, and says so rather than clamping to a slider end.

Off-slider values. When the implied number sits outside the range a slider covers, we refuse to clamp it — a clamped anchor is a wrong anchor. The tab falls back with the reason visible: The price implies ... far outside the range this slider covers. Showing preset inputs. You know exactly which tabs are anchored on the market and which are not.

There is also a currency and ADR gate on the strip: when the filed statements and the quote are in different currencies, or an ADR ratio makes per-share figures incomparable, the tiles come back empty with the reason instead of printing a P/E that is wrong by an exchange rate.

The same numbers in chat, MCP and the API

Everything above is a service layer, not a page, so the identical math is available three other ways.

In the AI chat on the site, ask for it in English: what does AAPL's price imply?

Over MCP or the REST API, four calls cover the whole surface:

• get_implied_assumptions("AAPL") — every method's implied parameter, each with solvable and, when false, the reason.
• get_valuation_parameters("AAPL") — the legal inputs, ranges, company defaults and filed growth presets.
• run_valuation("AAPL", anchor: "implied") — all seven methods with your overrides, each answer citing the filing it came from.
• save_valuation — persist a scenario, same store as the web history panel.

Free to start, credit packs from $10, no subscription — details on /developers. The same pages are on mobile.evidinvest.com.

Run it on something you actually hold

Pick one position you have an opinion about. Open its valuation page, read the implied EPS growth in the hero, then open its filed CAGR history and see whether the market is asking for the last decade repeated, doubled, or halved. That comparison takes about ninety seconds and it is the entire idea: a price is a set of assumptions, and you are allowed to disagree with them.

Research, not investment advice.