How We Calculate Fair Value
For about 40,000+ stocks worldwide we estimate a Fair Value, a company's intrinsic worth, from 26 independent valuation models, and we score business quality across 37 fundamental factors. This page explains transparently how that works, where the limits are, and why an estimate can be wrong.
We estimate what a business is worth from its own reported numbers, score its quality separately, and show how much evidence backs each estimate. Price is only the comparison, never the input.
- Annual reports
- 26 models
- Fair Value
- Quality Score
- Evidence
- Watchlist
- Portfolio
What "fair value" means
Fair value is an estimate of intrinsic worth from a company's fundamentals, independent of the current share price. We then compare it to the price: trading well below it, a stock looks undervalued by our model; above it, the valuation looks stretched. This is an analytical read, not a buy or sell recommendation.
How the fair value is built, step by step
Six stations sit between the filed accounts and the number you see. Every rule below exists verbatim in the calculation engine; none of it is simplified marketing.
Everything starts from the company's annual reports, never from the share price.
Future profits are worth less than today's. How much less depends on risk.
Every variant computes three scenarios: cautious (bear), base, optimistic (bull).
A revenue model unleashed on a bank produces nonsense. So not every company gets every model.
The families are blended with weights that depend on the company's growth profile (8 profiles).
Before you ever see the fair value, guards re-check every number at read time.
Only now comes the comparison: fair value versus price yields the upside. Valuation status, not a buy or sell recommendation.
Why our fair value differs from analyst price targets
The number on Yahoo or Investing.com is usually an analyst price target that hugs today's price. Our fair value is intrinsic worth from the fundamentals, built to diverge. That gap is the whole point.
Many users compare our fair value with the numbers on Yahoo Finance or Investing.com and wonder about the gap. The reason: these are two fundamentally different figures. The numbers shown there are usually analyst price targets, a 12-month forecast of where the price might go. Such targets almost always hug the current price and market sentiment and rarely stray far from it.
Our fair value, by contrast, estimates a company's intrinsic worth from its own fundamentals, independent of price and sentiment. It is deliberately built so it may diverge from the price: that divergence is the whole point, because it reveals where the market may be too expensive or too cheap. A fair value that always equals the price would be worthless. So a difference from other platforms is not an error, it is the product: an independent second opinion instead of the herd.
Two common follow-ups: "Then everything would be overvalued and you'd never buy." The opposite is true, the screener surfaces thousands of undervalued names; it is just honest enough to call many popular, expensive names expensive. And: even an excellent company can be too expensive. When a quality stock looks "overvalued" here, that does not mean "bad company", it means the price already bakes in growth and margins that still have to materialize. If you believe they will, you may value it differently, which is why we lay open the full breakdown and an evidence status for every stock.
The 26 valuation models
Each company is valued through a stack of independent intrinsic-value models, blended into one family-balanced consensus, weighted by how much trustworthy data backs it. The 8 model families, summarised in five groups:
- Discounted-cash-flow variants, future free cash flows discounted to today.
- Residual-income / earnings power, value from equity plus excess returns.
- Multiples, margin-anchored sales, earnings, book-value and EV multiples.
- Asset / book-value models, relevant for banks, insurers and holdings.
- Dividend models, Gordon growth and dividend DCF, only for reliable payers.
Not every company gets every model: banks and insurers, for example, only get book value, P/E, dividends and residual income. How many models a stock actually carries is shown on its stock page.
Every input is real reported data from the filings, nothing is guessed. Growth assumptions are deliberately capped so a single optimistic assumption can't dominate the result.
The Quality Score (37 factors)
Scores business quality from 0 to 100 using fundamentals only, independent of price, so a cheap price is never mistaken for a good business.
A separate layer scores business quality (0,100) from about 37 fundamental factors, profitability, growth, balance-sheet strength, returns on capital and more, independent of valuation. This keeps a cheap price from being mistaken for a good business: a low price with weak quality can be a value trap.
Evidence & confidence
Every estimate shows how much data backs it (high, medium or low) plus a bear, base and bull range, so the uncertainty stays visible.
Every estimate carries an evidence status (high / medium / low). "High evidence" means the fair value rests on rich, cross-verified data and enough applicable models, not a thin estimate from a few figures. Instead of a single point value we also show a bear / base / bull range so the uncertainty stays visible. The range is winsorised (bear at least 20%, bull at most 400% of the base value) so a single outlier model cannot stretch it into the absurd.
The backtest in brief
We simulated the strategy from 1992 to 2026, including companies from the USA and Europe that later failed. Result from 2001: 11.7% per year versus 8.6% for the MSCI ACWI Equal Weighted over the same days, after 0.25% ETF costs a year, a gap of +3.1 percentage points per year. This is the median over all 12 months in which the portfolio can be reviewed once a year, before taxes, dividends reinvested. Only stocks are tested (no funds, ETFs or trusts); the Quality Score uses the same six measures for every company. The Fair Values of the failed companies are estimated; the USA in the 1990s and Europe before about 2013 are incomplete.
A backtest is a simulation on historical data, the rules were chosen on the same history; past results are no guarantee of future outcomes. Rules, periods and risk are on the evidence page. How an honest test is built is explained in the handbook.
How the chart forecast is made
On the right of the "Price vs Fair Value" chart we show a five-year forecast. The number next to "Forecast" is the expected return per year, the most likely path, not a guarantee.
- The market as the starting point: what the overall market can deliver over the coming years at today's valuation, from our own index analysis and refreshed daily.
- The stock itself: its distance to fair value and a set of quality, earnings and financing signals. We only kept what held up in our tests across more than 30 years of market history, including years the model had not seen before.
The forecast is a hypothetical estimate. In our tests it separated well which stocks tended to do better and which ones lost money more often; single stocks can still deviate widely.
It is mainly about the path: the gap to fair value can close from both sides. The price can move toward the fair value, or the fair value can grow into the price. Our research in the handbook showed that strongly overvalued stocks often did not crash as much as expected, because their fair value kept rising.
The green line shows where the fair value is likely to head, judged by how comparable companies developed. The shaded area around the price line is the stock's usual volatility: single stocks often deviate clearly from the most likely path.
Limitations, why an estimate can be wrong
A fair value is an estimate, not a certainty. It can be off when:
- reported data contains errors or is later restated;
- one-off items distort the result;
- the business model, industry or interest-rate regime changes structurally;
- the conservative growth assumptions under- or over-state an exceptional company.
That is why the evidence status and range are part of every output.
Data source & freshness
Fundamentals, financial statements and prices come from a professional financial-data provider. Prices refresh every trading day after the close; fundamentals update as new filings arrive, fair values are recomputed daily. Every stock page shows the date its fair value is from. More under Data & Trust.
Frequently asked questions
Is the fair value a buy recommendation?
No. It is model-based, evidence-based analysis, not financial advice and not a buy or sell recommendation.
Does the fair value depend on the price?
No, it is computed independently from the fundamentals and only then compared to the price.
Why does your fair value differ from analyst price targets (e.g. Yahoo, Investing.com)?
Analyst price targets are 12-month price forecasts that hug the current price and market sentiment. Our fair value instead estimates intrinsic worth from a company's own fundamentals and is meant to diverge, that divergence is what reveals a possible mispricing. A difference is not an error but an independent second opinion. Even an excellent company can look expensive here when its price already prices in future growth.
How reliable is the estimate?
It depends on data quality and assumptions. Every stock therefore shows an evidence status and a bear/bull range.
Where does the data come from and how often is it recomputed?
Fundamentals and prices come from a professional financial-data provider; fair values are recomputed daily as new prices and filings arrive.
What backs the historical evidence?
A backtest from 1992 to 2026 including companies that later failed: from 2001 the strategy compounded at 11.7% per year before taxes, the MSCI ACWI Equal Weighted 8.6% over the same days, after 0.25% ETF costs a year (+3.1 percentage points; median of the 12 review months). Only stocks are tested, no funds, ETFs or trusts. Past results are no guarantee of future outcomes.