ValueScreener

Lesson 4 of 7

The average that is wrong

The arithmetic mean of ten real peer P/Es is 59.93 times. The harmonic mean is 26.70. One of them values a company at 961 billion and the other at 428.

Javier Sanz, founder

By Javier Sanz

Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.

Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)

In one minute

  • A P/E is a price divided by earnings, so averaging P/Es averages reciprocals and one near-zero denominator dominates the result.
  • On ten real peers the arithmetic mean was 59.93 times and the harmonic mean 26.70, and no company in the set traded near the first number.
  • The harmonic mean is what you want because it matches aggregating the set: total value over total earnings.

This is the one piece of arithmetic in relative valuation that is simply wrong in most spreadsheets, and it is wrong in a direction that flatters.

Average a real peer set four ways

  1. Take the set. Ten US-listed household and personal products companies above 2 billion dollars of market value with a computed ratio, on prices last stored 2026-09-15: 21.73, 32.95, 192.50, 23.27, 16.20, 31.10, 35.15, 13.29, 212.16 and 20.95 times.
  2. Arithmetic mean. Add them and divide by ten: 59.93 times. Eight of the ten companies trade below 36 times, so the mean describes none of them.
  3. Median. Sort and take the middle pair: 27.19 times. The two extremes stop mattering.
  4. Harmonic mean. Average the reciprocals and invert: ten divided by the sum of one over each multiple is 26.70 times.
  5. Aggregate. Add every company's market value and divide by the sum of their earnings: 576.0 billion over 23.93 billion is 24.07 times. This is the figure the harmonic mean approximates, and the median lands near it too.

Four averages of the same ten multiples

  • Arithmetic mean59.93x
  • Median27.19x
  • Harmonic mean26.70x
  • Total value over total earnings24.07x
Ten US household and personal products companies above 2bn dollars of market value, prices last stored 2026-09-15.Source: The product's own company catalogue, read 2026-09-16; ratios as published on prices last stored 2026-09-15.

Why the arithmetic mean fails

A multiple is a ratio with earnings underneath. A company earning almost nothing produces an enormous multiple, and an enormous number dominates a sum.

Two of the ten companies here carry a net margin near 1 percent, which puts them at 192.50 and 212.16 times. Those two alone contribute 40.5 of the 59.93 point arithmetic mean. The average is a statement about two poor years, not about the industry.

The harmonic mean inverts first. In reciprocal space a company earning almost nothing contributes almost nothing, which is the correct weight: it has little earnings power to average.

What it is worth in dollars

The arithmetic mean values this business at more than twice what the harmonic mean does. Nothing separates the two figures except which average was typed into a cell.

The rule, and the check

Use the harmonic mean, or equivalently compute the aggregate: add the market values, add the earnings, divide. Report the median beside it. When those two disagree by much, the set is being driven by a few members and you should say so rather than picking the one you prefer.

The same problem runs through every multiple with a small or volatile denominator: enterprise value to EBITDA, price to free cash flow, price to book where book equity is close to zero. It does not affect yields, because a yield is already the reciprocal: earnings yields average correctly with an arithmetic mean.

Check yourself

Why does the arithmetic mean of a set of P/Es overstate the group?

  1. Because P/Es are usually too high
  2. Because a company with almost no earnings produces an enormous ratio that dominates the sum
  3. Because it ignores company size

Show the answer

Answer: B.

The denominator is earnings. Two poor years in a ten-company set contributed 40.5 points of a 59.93 point mean.

What does the harmonic mean of a peer set's P/Es correspond to?

  1. The median company
  2. Aggregating the set: total market value over total earnings
  3. The lowest multiple in the set

Show the answer

Answer: B.

Inverting first, averaging, and inverting back is the same operation as adding the values and the earnings separately, which is why the two land close together.

Which figures can be averaged with an ordinary arithmetic mean without this problem?

  1. Price to book ratios
  2. Yields, such as an earnings yield, because they are already reciprocals
  3. Enterprise value to EBITDA multiples

Show the answer

Answer: B.

A yield puts the volatile figure on top, so a company earning almost nothing contributes a value near zero rather than a value near infinity.

Sources

  • The product's company catalogue, read 2026-09-16: ten US-listed household and personal products companies above 2 billion dollars of market value, with ratios computed on prices last stored 2026-09-15.
  • Damodaran, Investment Valuation: the harmonic mean as the correct average for a multiple, and the aggregate ratio it corresponds to.
  • Mauboussin and Rappaport, Expectations Investing, 2001: reading a set of multiples as a set of embedded expectations rather than as a population to average.

What this leaves out

  • The peer set here is one industry on one date and is used to demonstrate the arithmetic, not to value anything.
  • Earnings behind each multiple are implied from the published ratio and market value rather than read line by line from each filing.
  • Ratios and prices were current on 2026-09-15 and move every session.
  • Nothing here is advice about any security.

The model, not your inputs

Lesson 4 of 7 of Multiples, and why they mislead. It teaches how this part of a valuation is built. The inputs are still yours, the page supplies none of them, and arithmetic on invented numbers is a worked example rather than a value.

Free audit

Research tool, not investment advice. See our methodology for how the product reads a filing and what it refuses to compute.

Once a week, what changed in the filings of US-listed companies, why it can matter to a thesis, and what deserves a closer review. Grounded in filings, not market noise.