How do you choose comparable companies for a comp set?

Q: When you value an orange farm, should you include an apple farm in the comps?

It depends, and the interviewer is testing whether you can back that up. First ask whether you have enough other comps to leave the apple farm out. If you do, leave it out. If the peer set is thin, an apple farm may be the closest available comparable, so include it and state the overlap: same industry, same business model, possibly comparable size, possibly the same part of the world. Close by saying it is the best data available to produce the best possible comparable estimate.

Q: How do you choose comparable companies for a trading comps analysis?

Screen on industry, business model, size, geography, and growth and margin profile, and require that the company is publicly traded so there is a live market price to pull. What you are building is a set that holds up as a whole and can carry the multiple you plan to run on it, whether that is EV/EBITDA, EV/Revenue, or P/E.

Q: Why should the target company never be included in its own comp set?

Leaving the target in its own peer group drags the median toward the target itself, so the analysis quietly becomes a comparison of the company to the company. It is the first check to run when an interviewer hands you a comp set and asks what is wrong with it.

Q: Do you use basic or fully diluted share count in a comp set?

Fully diluted. Options, restricted stock, and convertibles are real claims on the equity. Using basic shares understates the share count and therefore understates the equity value behind every equity multiple in the set, P/E included. Diluted shares are what make the equity value right.

Q: What are the most common errors in a comparable companies analysis?

Four tend to be hiding in there: the target sitting inside its own comp set and dragging the median, basic share count used where fully diluted is required, mixed time periods across the peers, and one-time items left in the earnings you divide by. Catching them is mechanical. Know what a proper comp set looks like, then check the details one by one in the same order every time.

How to pick and screen a trading comps peer set, defend a borderline comparable, and catch the four errors usually hiding in a comp analysis.

OFFERGOBLIN·6 min read·updated August 2026

On this page

The orange farm question
Why the peer set does most of the work
How to back up "it depends"
Ask whether you need the comp at all
Say what actually overlaps
Close on the standard
The four errors usually hiding in there
The target is sitting in its own comp set
Basic share count instead of fully diluted
Mixed time periods
One-time items left in
How you catch them
Interview script

"It has been demonstrated time and time again that no asset is so good that it can't become a bad investment if bought at too high a price." (Howard Marks)

The orange farm question

At some point in a valuation interview you get a version of this one: you are valuing an orange farm. Would you include an apple farm in the comps?

Answering yes or no is the trap. Both answers are defensible, and the second you commit to one, the interviewer asks why. The question tests whether you can hold a judgment call and argue for it out loud.

So you give the classic consulting answer. It depends. On its own that is worth nothing. What earns the point is the part that comes after it.

Why the peer set does most of the work

A house does not get priced by forecasting its future cash flow or adding up the cost of its materials. Somebody looks at the neighborhood, looks at what other houses sold for, and compares on a square foot basis. Trading comps do the same thing to businesses. Find companies like this one, pull their multiples, see where yours lands.

That makes the peer set the analysis. A company trading at 12x earnings has no verdict attached to it until you know who else is in the room. Change the room and the verdict flips. Every multiple you quote is a claim about one business relative to the neighbors you picked for it.

Interviewers know this, which is why comp selection is where they go hunting. It shows up two ways. They push on a borderline peer and make you justify it, which is the orange farm. Or they hand you a finished comp analysis and ask what is wrong with it, because the same handful of errors tends to be hiding in there.

How to back up "it depends"

Ask whether you need the comp at all

Before you defend the apple farm, ask how deep the bench is. Do we have enough other comps that we do not need this one? If yes, throw it away. Nobody has to litigate a marginal peer that the set can live without.

If the bench is thin, the calculus changes. An apple farm may be the closest thing to an orange farm you are going to get. In that case it goes in, and you say out loud why it is in there.

Say what actually overlaps

Then argue the similarity on specifics. Two farms growing different fruit still share a lot: same industry, same business model, possibly comparable size, possibly the same part of the world. Those are the screens, and they are the same screens whether the target is a citrus grower or a payments company.

Screen	The question you are answering
Industry	Does it sit in the same industry?
Business model	Does it make money the same way?
Size	Is it in the same size range?
Geography	Does it operate in the same part of the world?
Growth and margin	Does it grow at a similar rate and earn a similar margin?
Publicly traded	Is there a live market price to pull? Trading comps only work on public companies.

What you are building is a set that holds up as a whole, and that can carry the multiple you plan to run on it, whether that is EV/EBITDA, EV/Revenue, or P/E.

Close on the standard

Finish with the sentence that ends the argument: this is the best data available to give us the best possible comparable estimate for an orange farm. That line tells the interviewer you understand what a comp set is. It is a judgment call, it produces an estimate, and you are willing to own both.

There is never a clear answer with comps. It is more about your ability to defend and explain the one you picked.

The four errors usually hiding in there

When an interviewer hands you a comp analysis and asks what is wrong with it, these are the ones to look for.

The target is sitting in its own comp set

The company you are valuing should never be in its own peer group. Leave it in and it drags the median toward itself, so the analysis quietly becomes a comparison of the company to the company.

Equity value built on basic shares is wrong. Options, restricted stock, and convertibles are real claims on the equity, and leaving them out understates the share count and therefore the equity value behind every equity multiple in the set, P/E included. Use the fully diluted count so the equity value is actually right.

Mixed time periods

Every company in the set has to be on the same clock. Pick a period, the same quarter or the same year, and make every comp fall inside it. A set that puts one company's latest quarter next to another company's full fiscal year is not measuring the same thing twice.

One-time items left in

Businesses take one-off hits: a lawsuit, a settlement, a disaster at a facility. Leave those in the earnings you divide by and they skew the multiple, sometimes badly. Strip out what is not going to repeat.

How you catch them

Catching these is mechanical. It requires you to know what a proper comp set looks like and then be thorough, checking the details one by one. Run the same pass every time:

Is the target sitting in its own set? Take it out.
Are the share counts fully diluted?
Is every company on the same period?
Has anything non-recurring been stripped out of the earnings?

Four questions, asked in the same order, on a page you have never seen before.

Interview script

It depends, and here is what it depends on. First I would ask whether we have enough other comps to leave the apple farm out. If we do, leave it out. If we do not, an apple farm is probably the closest thing to an orange farm available: same industry, same business model, likely similar size, possibly the same part of the world. I would include it and say plainly that this is the best data available to give us the best possible comparable estimate, and I would defend it on those grounds.

create your free account to keep reading

11 more sections. free forever. takes 30 seconds.

By continuing, you accept our Privacy Policy.

Frequently Asked Questions

When you value an orange farm, should you include an apple farm in the comps?

How do you choose comparable companies for a trading comps analysis?

Why should the target company never be included in its own comp set?

Do you use basic or fully diluted share count in a comp set?

What are the most common errors in a comparable companies analysis?