Why fair value is a range, not a point

A fair value estimate looks precise: a single number like $142.30 on a screen. In reality it is the output of a chain of assumptions, and small changes in those assumptions move the answer a lot. A more honest way to read any valuation is as a range. Understanding why helps you avoid false precision, which is one of the most common errors in using valuation tools.

Where the uncertainty comes from

Every valuation method depends on inputs that nobody can know:

• Future growth. Revenue and profit several years out depend on competition, regulation, technology and the economy.
• Margins and reinvestment. How profitable the business will be, and how much capital it will need, are both forecasts.
• The discount rate. It is estimated from a risk-free rate, an equity risk premium and a beta, and each of those is itself uncertain. See discount rate and CAPM and equity risk premium.
• The terminal value. In a discounted cash flow, most of the value often sits beyond the explicit forecast period. See terminal value explained.
• Peer multiples. They are themselves market prices, so they move with sentiment.
• Accounting. Reported earnings, depreciation and one-off items can all be defined in different ways.

A quick sensitivity check

Take a company expected to generate $100 million of free cash flow next year, growing 3% a year forever. The value is that cash flow divided by the gap between the discount rate and growth:

Value = FCF next year / (r - g)

A change of one percentage point in the discount rate, well within normal disagreement, moves the value by roughly 15 to 20 percent. Move the growth rate by a point as well and the spread widens further. Two careful analysts can defend very different numbers without either being wrong.

Why combine several models

Different methods fail in different ways. A peer-multiple model reflects the market's current pricing, a DCF reflects cash flow forecasts, and a regression-based multiple adjusts for quality. Blending them reduces the damage from any one model's blind spot, although it cannot remove the shared uncertainty. The related articles on relative valuation with peer multiples and earnings power value describe two of the building blocks.

How StocksXray handles it

StocksXray's fair value is an equal-weight blend of three models: peer multiples (sector median EV/EBIT, P/E and P/FCF), a fundamentals-adjusted multiple (a regression on growth, ROIC and margin), and a 10-year DCF scaled to the market's average. Earnings power value is shown separately as a floor and is not blended in.

The site is explicit that typical model error is around 40%. Reflecting that, the labels use a margin that depends on how uncertain the models are for a given company: 20%, 30% or 40% for Low, Medium or High model uncertainty. In broad terms, a stock is labeled Undervalued when its price sits below fair value by more than that margin, Overvalued when it sits above fair value by more than that margin, and Fair when it falls in between. The less certain the models are for a company, the wider the margin, and the larger the gap between price and fair value must be before the label changes.

Example (illustrative numbers)

Suppose the blended fair value of a stock is $100 and its model uncertainty is Medium, so the margin is 30%. Roughly, the stock would count as Fair for prices between about $70 and $130. At $60 it would be Undervalued, and at $140 it would be Overvalued. Under High uncertainty the margin of 40% would widen the Fair band to about $60 to $140, and under Low uncertainty the 20% margin would narrow it to about $80 to $120. The same price can therefore earn a different label depending on how confident the models are.

How to read it

Rules of thumb:

• Treat the label as a way to sort ideas, not a trade signal. A stock labeled Undervalued has been flagged as worth a closer look.
• The further the price sits from the range, the more it takes to explain the gap by estimation error alone.
• A wider margin is the model telling you that the estimate is shaky.
• Compare the estimate with your own view of the business. If you cannot explain why the market disagrees, treat the signal with caution.

Common mistakes

• Confusing precision with accuracy. Showing a number to two decimals does not make it accurate.
• Buying on a small discount. A 5% gap between price and fair value is far inside the noise.
• Anchoring on one model. If the models disagree widely, that disagreement is information.
• Treating fair value as a target. Prices can stay above or below fair value for years.
• Skipping a margin of safety. The traditional answer to uncertainty is to require a large discount before acting. See margin of safety and intrinsic value vs market price.

You can browse the stocks currently labeled Undervalued, Fair and Overvalued, and read the full approach on the methodology page.

Key takeaways

• Fair value rests on forecasts and rates that cannot be known, so it is better read as a range.
• Small changes in the discount rate or growth can move a valuation by 15 to 20 percent or more.
• StocksXray blends three models, shows earnings power value as a floor, and widens its margin as model uncertainty rises.
• Use the label to prioritize research, then verify the business yourself.