The 3 Claude prompts I use to find undervalued companies

Cheap stocks are everywhere.

Stocks that are cheap for the right reasons are not…

That is the whole game.

And it is harder than it sounds.

These are the three prompts I use when I am hunting for real undervalaution.

Not the kind that shows up on a screener.

The kind the market has not figured out yet

I use Claude Sonnet 4.6. Free version works for all of these.

Claude is only as useful as the instructions you give it.

A vague prompt gets you a Wikipedia summary.

A structured prompt gets you something that looks like it came from a buy-side analyst who has done this before..

The difference is specificity.

Tell Claude exactly what role to play, what to produce and how to format it.

The output quality jumps dramaticaly.

These three prompts do that.

Some companies trade cheap because they are bad businesses.

Others trade cheap because they are hard to understand.

This prompt helps you figure out which one you are dealing with.

I use it when a valuation looks low but I cannot immediately explain why the market is wrong.

It does force Claude to reconstruct the bull case from first principles instead of just repeating consensus.

Why it works: Claude is very good at separating genuine business problems from perception problems. That distinction is where most of the opportunity in value investing actually lives.

The prompt:

Paste in a recent earnings call transcript or investor presentation after the prompt. The more context you give, the sharper the output.

Relative valuation only works if you are comparing the right companies.

Most screeners build peer groups by sector code. That produces lazy comparisons. A regional bank gets benchmarked against a global investment bank. A niche software business gets compared to enterprise players with totally different growth profiles.

This prompt builds a better peer group and uses it to figure out whether a discount is justified or a mispricing.

Why it works: Claude forces you to be precise about what makes a company truly comparable. That discipline alone changes how you think about valuation.

The prompt:

This one works best when you already have some numbers in front of you. Give Claude the actual figures and it builds around real data, not estimates.

A cheap stock can stay cheap for a long time.

This is the part most value investors skip.

They find the gap, build the thesis and then wait.

Sometimes for years.

The missing piece is identifying what will actually cause the market to reprice the stock.

This prompt forces that question directly. It is uncomfortable.

It is supposed to be.

Why it works: It separates genuine value from value traps.

Most investors do not want to answer the hard questions at the end of this prompt.

That is exactly why they matter.

The prompt:

Use them in sequence.

Start with the misunderstood business decoder to understand why the market might be wrong,

Then run the comparable company pressure test to quantify the gap and whether it holds up….

Then finish with the catalyst identifier to understand what needs to happen for the thesis to pay off and how long you might be waiting.

Run all three and you have covered more ground than most analysts do in a week. In about an hour.

They work today.

Pick a company you already think looks cheap and run it through all three this week. You might confirm your thesis. You might find the reason everyone else passed.

Either outcome is worth knowing.

Send this to someone who is still using screeners as their only research tool.

Check out my full Claude Investing Prompt Library and get 10 additional proven prompts to sharpen your research workflow.