> ## Documentation Index > Fetch the complete documentation index at: https://docs.chicago.global/llms.txt > Use this file to discover all available pages before exploring further. # Bottom-Up Beta > Understanding bottom-up beta - estimating a company’s systematic risk from comparable businesses rather than from its own share-price regression A bottom-up beta estimates a company's sensitivity to market risk from the operating risk of comparable businesses, instead of regressing its own share price against an index. ## Beginner ### What It Means The usual way to measure beta is to look at how a stock has moved against the market. That works when the stock trades actively and its history is representative. When a share trades thinly, or the listed entity is a poor proxy for the underlying business, the regression produces a number with more noise than signal. The alternative is to look at companies doing the same thing, strip out the effect of how much each of them borrowed, take a stable central estimate, then add back the target company's own borrowing. ### Example A newly listed regional operator has 14 months of thin trading history. Rather than trust that regression, you take a set of established operators in the same business, remove their leverage, take the central estimate, and re-apply the target's own capital structure. ### Why It Matters Beta feeds directly into the discount rate, and the discount rate moves valuation substantially. Illiquidity can distort an estimated beta and affect the resulting discount rate and valuation. *** ## Advanced ### How to Read It `Unlevered beta = levered beta / [1 + (1 − tax rate) × debt/equity]` `Relevered beta = operating beta × [1 + (1 − tax rate) × target debt/equity]` Both steps use market-value debt and equity, and both assume the debt carries no systematic risk of its own. Where leverage is high or the credit is distressed, that assumption understates asset risk. A bottom-up estimate is not automatically superior to a regression. A global peer set can understate local market segmentation, regulation, or funding risk that a local regression genuinely captures. Both estimates should be shown, and a large gap between them explained rather than averaged. Two related corrections address measurement noise in a regression beta. Lead-lag adjustment handles nonsynchronous trading, where a thinly traded share reacts to market moves a day late. Statistical shrinkage pulls an imprecise extreme estimate toward a central value. ### Common Misreadings * **Assuming an industry beta is conservative**: in a large cross-section, substituting one can lower the discount rate rather than raise it. * **Reporting beta without its standard error**: a single-stock beta is an estimate with sampling error, not a fixed company characteristic. * **Ignoring the index choice**: beta against a local index and against a global index answer different questions. ### In Parallax Reports Parallax reports the observed company beta, the operating-beta diagnostic and the price-implied discount rate as three distinct pieces of evidence, alongside trading-frequency and explanatory-power statistics. See [Cost of Capital](/methodology/valuation/cost-of-capital). ### Related Terms The underlying concept Where beta enters the valuation How co-movement is measured