A Multiple Has No Verdict Until a Peer Group Supplies the Reference Point

The absolute value of a multiple contains no conclusion, because a multiple is a fraction in which the numerator was set by the market and the denominator was computed by one company's accountants, under their own accounting policy, at their own point in the cycle. Comparison against an industry turns that fraction into a judgement: it supplies the reference point a lone number simply does not have. But the comparison only works when the group is assembled around the mechanism that generates profit, rather than around a formal industry code — otherwise you are not comparing companies, you are comparing different ways of recognising revenue and different tax regimes.

What is actually being compared when you "compare with the industry"

A multiple answers one question: how much the market pays for a unit of whatever the company produces for its shareholder — earnings, revenue, book equity, operating cash flow. The question "expensive or cheap" has no content without the second half of the comparison. The price of a unit of earnings is neither good nor bad in itself; it is either above or below what the market charges for a comparable unit at similar businesses.

The easiest way to see this is in practice. Put 3,77 next to 45,75 — the figures sit on different scales, and nothing follows from that except that a bank and an oil company earn money in different ways. For a bank, debt is raw material rather than a burden, which is why it is valued through equity and return on equity. For an extraction business, what matters is reserves, lifting costs and a tax regime changed by government decisions rather than by management. Lining companies up across industries is not a rigorous comparison; it is an optical illusion. The mechanics of the fraction itself are covered separately: why a low P/E does not mean cheap.

How to assemble the comparison group

The peer group is the central decision in this work, and no amount of computational precision will repair an error made there. The criteria under which companies are genuinely comparable:

• The source of revenue, not the industry label. A company may be classified as a retailer while actually earning its money from logistics and financial services — and then its denominator is built differently from those of its sector neighbours.
• Capital structure. Multiples based on earnings per share are sensitive to debt: borrowed money reduces profit through interest expense, so a heavily indebted company looks "more expensive" at identical operating efficiency. Multiples built on enterprise value are nearly neutral to this, which is why capital-intensive industries are usually read through them.
• Phase of the cycle and capital expenditure. A company at the peak of an investment programme and a company that has finished one produce incomparable denominators even inside the same industry.
• Accounting policy. Asset revaluations, capitalised costs, lease recognition and currency translation differences all move reported profit without moving the business. Hence the practical rule: check your conclusions against the cash flow statement, because cash is harder to fake than profit.
• Comparable disclosure. If part of the group has stopped publishing segment reporting, the group is skewed — what survives is the data of those who benefit from showing it.

It is more convenient to verify the group's composition against the general list of Russian equities and against publication dates in the earnings calendar: multiple values jump on the day a report lands, and half a group still measured on the old period yields a biased median.

The median, not the average

Industry samples on the Russian market are small, and an average across them breaks on a single outlier — a company, say, whose one-off asset revaluation pushed profit close to zero and inflated the fraction to a meaningless size. The median withstands such cases. More useful than the median is the dispersion inside the group: a narrow spread says the market regards these businesses as homogeneous, a wide one means the market sees different stories within the same industry — and then the "industry norm" is an artefact of averaging rather than a benchmark.

An entire industry can also be mispriced

The method's central trap: comparison with an industry tells you where a company sits inside the group, and nothing about whether the group is correctly priced. If a whole sector is cheap relative to its own history, there is usually a reason — regulatory pressure, a falling price for the underlying commodity, the loss of export markets, deferred dividends. A share trading below the median of a dying sector remains a share in a dying sector.

A separate factor is the policy rate. The required return sets the ceiling for every multiple in the market simultaneously, and when it moves, an entire industry compresses or expands at once. That is why industry comparison is read alongside the regulator's wording and, in the bond part of a portfolio, the mechanics of duration. The alternative to a share is not another share but the risk-free yield on OFZ.

Russian specifics

Here the method carries limitations that should not be passed over in silence. The industries are narrow: in some sectors there are almost no listed comparables at all, and the "industry median" ends up computed across a group in which every company is its own sector. Free float is modest at many issuers, which means the numerator of the fraction is formed on a thin market — why that matters is covered in the piece on liquidity. Finally, state participation changes the logic of profit distribution: dividend policy becomes a fiscal decision, so the thing to track is actual announcements in the payout calendar — the nearest ones being {{dividend_calendar|limit=5}}.

The order of work

First, determine what you are paying for in this particular business and choose the denominator to match it, rather than out of habit. Second, assemble the group by revenue mechanism and capital structure. Third, bring every value onto the same reporting period. Fourth, look at the median and the dispersion, then go looking for the reason a specific security deviates. Fifth, check whether the group as a whole is priced anomalously against its own history and the current rate.

An individual instrument card with its set of indicators looks like this — {{instrument:MGNT}}; definitions of the denominators are collected in the glossary.

What industry comparison does not give you: it does not forecast future earnings, it does not account for management quality, and it does not replace reading the financial statements. It is a tool for posing a question — where does this company trade differently from those that resemble it — and not a tool for answering one.