Methodology · Valuation dispersion
Valuation dispersion map
A cross-sectional plot of valuation multiples against quality and growth used to characterize whether the market is pricing a narrow leadership cohort or paying for fundamentals broadly.
Specification
Valuation dispersion map — operational spec
Quantifies the spread of valuation multiples across the index. Compressed dispersion is a late-cycle marker; decoupling between valuation and quality is the model's most important alarm.
Inputs
- Forward P/E, EV/EBIT, and P/B for every constituent (consensus, monthly snapshot)
- Trailing 5-year mean ROIC for every constituent
- 3-year forward EPS growth rate (consensus, monthly)
- Index membership and float-adjusted weights at the snapshot date
Computation
- For each multiple, compute the 10th / 50th / 90th percentile across constituents.
- Dispersion = 90th percentile minus 10th percentile, expressed as a multiple of the median.
- The dispersion metric is converted to a percentile rank against its own multi-decade history.
- Each constituent is plotted on a (valuation multiple) × (ROIC) plane and a (valuation multiple) × (forward growth) plane.
- A decoupling flag is raised when the highest-multiple cohort no longer overlaps with the highest-ROIC cohort.
Outputs
- Dispersion percentile rank (0-100) for forward P/E, EV/EBIT, and P/B.
- Two scatter plots: valuation × quality and valuation × growth.
- Decoupling flag (boolean) and the magnitude of the decoupling.
Limitations
- Index composition changes over time; the percentile rank silently incorporates survivorship bias.
- Forward growth and forward P/E inherit sell-side estimate noise.
- REITs, banks, and other capital-structure-distinct cohorts distort the index-wide read; the model publishes a financials-and-REITs-stripped variant alongside the headline.