Comparable company analysis, known everywhere as comps, values a company by comparing it with similar businesses using valuation multiples. If a group of peers trades at a typical multiple of earnings or cash flow, applying that multiple to the company in front of you gives an implied value. The method is fast and anchored in what investors are actually paying today, and it lives or dies entirely on the quality of the peer set.

This module covers how comps work, why choosing the peers is the whole game, how a range rather than a point is built from a peer set, and the pitfalls. Chief among those is the flattering peer group that quietly inflates a target without appearing to argue for anything.

What comps do

It helps to see comps next to the DCF you met in the last module, because the two methods work from opposite directions.

A DCF builds a value up from the company’s own cash flows and pays no attention to what anyone is currently paying for the shares. Comps do the reverse. They borrow the value from the market. The reasoning is that if similar businesses trade at, say, 20 times EBITDA, then a business much like them ought to trade somewhere near 20 times too.

What you get out is a market anchored value: an estimate of what investors are willing to pay today for a company of this kind. That is a different question from what the business is worth, and knowing which question you have answered is the beginning of using comps properly. If the whole sector is expensively priced, comps will tell you the company is fairly valued, because comps have no way of noticing that the sector is expensive. A DCF might. This is precisely why analysts run both.

Choosing the peer set: the make or break step

Everything downstream depends on this one decision, which is why it is also the most abused step in the method.

A genuine comparable shares the things that actually drive valuation, which are sector, business model, size, growth rate and margins. Note that “same industry” is nowhere near sufficient on its own. Two software companies in the same market can deserve very different multiples if one grows at 25% and the other at 8%, because the buyer of the faster one is paying for more future revenue.

Take a concrete case. A £400M cybersecurity company growing 25% with 72% gross margins belongs with mid cap, high margin software peers of a similar shape. It does not belong with a £4.5B platform growing 18%, however famous that larger name is and however satisfying it is to appear beside it.

What makes this trick so effective is that it never states a falsehood. Every company on the list is real, every multiple is accurate, and the arithmetic is correct. The claim being smuggled through is the unstated one: that these companies belong in the same group. Because it is never written down, it is never defended, and a reader who checks the sums will find nothing wrong.

So when you read a comps table, read the list of names before you read the numbers. Ask of each one whether you would swap it for the subject company at the same price. Any name that gives you pause does not belong in the median.

The multiples used

Different multiples suit different situations, and a serious analysis will usually run more than one, because agreement between them is itself evidence.

Choosing the wrong one produces a figure that looks precise and means nothing. A P/E tells you very little about a loss making company, and EV/Revenue used on a mature business ignores the fact that profitability is the whole question at that stage. Worse still is mixing multiples across companies that differ in how much debt they carry, since EV based multiples and price based multiples are measuring different slices of the same business.

One rule keeps you out of most trouble. Whatever multiple you use, make sure the numerator and the denominator refer to the same thing. Enterprise value pairs with EBITDA and with revenue, because all three describe the whole business. Share price pairs with earnings per share, because both describe the shareholders’ portion after the lenders are paid. Cross the two and the comparison is meaningless.

From multiple to value

The mechanics are the easy part. Take the peer group’s representative multiple, apply it to the subject company’s own metric, and you have an implied value.

So a peer median of 20× EBITDA applied to the company’s EBITDA gives an implied enterprise value. Subtract net debt to reach equity value, then divide by the share count for a per share figure. That final step is the same one you met in the DCF module, and for the same reason: lenders are paid before shareholders.

The median is used rather than the average, and the reason matters. An average is dragged by extremes, so one peer trading at 60× because it is the subject of takeover rumours would haul the whole valuation upwards. A median ignores how far away the outlier sits and only counts that it is above the middle. With a peer set of eight or ten companies, one strange name is common, so the median is the safer measure of the group’s centre.

A range from the peer set

A peer set does not really produce one multiple. It produces a scatter of them, and collapsing that scatter to a single figure throws away the most honest thing the method has to offer.

So analysts often map the spread onto scenarios using percentiles: the 25th percentile as a bear case, the median as the base case, and the 75th percentile as a bull case.

This is worth more than a tidier presentation. It states plainly that the market itself does not agree on what companies like this are worth, and it shows you the width of that disagreement. A peer set clustered tightly is telling you the market has a settled view, so a value derived from it carries some weight. A peer set scattered from 8× to 40× is telling you there is no consensus at all, and any single number drawn from it is close to arbitrary. The spread is information about how much to trust the answer.

Why a discount to peers is not automatically cheap

Here is the error that costs beginners the most money, and it follows naturally from everything above.

Having built a peer set and found the subject trading below the median, the temptation is to read that gap as an opportunity. Often it is not. A company can trade below its peers for entirely sound reasons: it grows more slowly, its margins are thinner, it carries more debt, or its business is simply riskier. Higher growth genuinely deserves a higher multiple, because the buyer is purchasing more future earnings, so a lower multiple on a slower company can be exactly correct.

Put the question the right way round. Not “is it cheaper than its peers?” but “is the discount deserved?” To answer that you have to go back to module 2 and compare the dashboards: the growth rates, the margins, the returns and the leverage of the subject against the peers. If the subject matches them on the fundamentals and still trades lower, you may have found something. If it lags them on every measure, the discount is the market pricing the difference, and the shares are cheap for a reason.

Blending comps with the DCF

Comps and a DCF answer the same question from opposite directions, one from the market and one from the cash flows, so analysts often combine them into a blended target, weighting each method according to how much they trust it in this particular case. The arithmetic is simply blended target = (DCF × weight) + (comps × weight). If a DCF gives £22 and comps give £25, a blend lands somewhere around £23 to £24.

The weights are a judgement, and they should be explained rather than assumed. An analyst might lean on the DCF for a mature business with predictable cash flows, and lean on comps for a young company whose forecasts are guesswork but whose peer group is well defined.

The most valuable thing about running both is not the blended number at all. It is what happens when the two disagree. Close agreement raises your confidence, because two methods with different weaknesses arrived at the same place. A wide gap is a question that has to be answered before either figure is usable. If the DCF says £40 and comps say £22, one of two things is true: the forecast is more optimistic than the market’s, or the market is missing something the forecast has seen. Both are interesting, and neither is settled by averaging them.