Rethinking P/E Ratios in Small-Cap Growth: A Practical Test of a Better Metric

Why the familiar P/E ratio often misleads and how a simple alternative, S=(P−E)/P, opens the door to sounder portfolio analysis

Rethinking P/E Ratios: A Small-Cap Growth Case Study

Introductory note: This post is part of my valuation metrics series.

For a limited time, I’m offering a 50% discount on paid memberships. Everything up to the paywall is free; the cleaned dataset, detailed tables, and test outputs appear under the paywall.

Introduction: Why This Matters

The price-to-earnings (P/E) ratio is the most widely quoted valuation measure in markets, yet it behaves badly in exactly the kinds of companies that dominate small-cap growth portfolios.

This post does three things:

1. Explains the five persistent problems with using P/E in portfolio analysis.
2. Introduces an alternative statistic, S = (P - E)/P, which is defined for every company — even those with zero or negative earnings.
3. Illustrates the contrast using the top-20 holdings of a leading small-cap growth fund.

The detailed results — including the cleaned dataset, statistical tests, and interpretations — are available to paid members below the paywall.

The Five Problems with P/E

1. Excluding Firms with Negative Earnings

P/E is undefined for loss-makers and is often reported as “N/M.”
Databases and analysts typically drop these firms, which makes average P/E look cheaper than it really is.

2. Instability around Near-Zero Earnings

A tiny change in earnings near zero can make P/E jump or collapse, or become undefined.

3. Distortion of Averages and Skewness

A few firms with very high P/Es (often due to tiny but positive earnings) can pull up the mean and distort the distribution.

4. Misleading Portfolio Summaries

A single portfolio-level “median P/E” can hide the fact that many constituents are unprofitable or that a few firms with extreme multiples dominate the headline figure.

5. Statistical Inference Headaches
Skewness, heavy tails, and missing observations mean that conventional t-tests and even non-parametric tests often can’t be applied consistently.

The Alternative: S = (P - E)/P

- Defined for all firms: even if earnings are negative.

- Valuation signal: for profitable firms, S = 1 - (E/P); lower earnings yield → higher S → higher valuation.

- Stable near zero earnings: tends to 1 rather than exploding. Reduces impact of a small number of extreme observations on averages.

- Comparability: means, medians, percentiles can be computed without excluding loss-makers.

- Enables inference: a continuous, finite variable that can be transformed or tested with both parametric and non-parametric methods.

Case Study Overview

We applied this idea to the 20 largest holdings of I-shares Morningstar small-cap fund ISCG.

Key observations:

- Four firms had negative EPS → undefined P/E.
- Using S, all 20 firms are included.
- P/E showed extreme skewness and variance; S was tight and stable, centered just below 1.

The next section (for paid members) presents the full cleaned dataset, descriptive statistics, hypothesis tests, and normality check.

Get an annual subscription 50 percent, off only $30.

https://bernsteinbook1958.substack.com/4d9daaf9

Detailed Results for Paid Members

The 20 largest holdings of fund ISCG.