I cloned how a $77B investor thinks in Claude Code

Chris Hohn just did something he rarely does. He talked.

Hohn runs TCI, a London fund most retail investors have never heard of.

Last year his fund made investors $18.9B. That is the largest one-year gain any hedge fund has ever posted.

Then he sat for a rare, long-form interview, and the Financial Times got hold of his latest letter to investors.

Buried in that letter was the move that surprised me. He cut roughly $8B of Microsoft, a position he had held since 2017.

His reason was not valuation. It was that AI might break the moat.

So here is a man who just printed a record, telling us how he is repositioning for AI.

The problem is obvious. How do you actually learn how a mind like that works?

And how do you know your own portfolio isn’t sitting on the wrong side of his call?

I wanted to use AI to reverse-engineer exactly how Hohn thinks, then stress-test a portfolio against it.

Here’s the plan:

• Pull his interview and his letter
• Build a profile of how he thinks
• Score a portfolio against his lens
• Find the collisions

The Prompt

I ran this in Claude Code, in plan mode, so it interviews me before it builds anything.

That last part matters. The model asks me questions first, which kills the gaps before they become a bad output.

If parts of this prompt fit your own process better another way, adjust the parameters.

If you want to go further, you can connect a live market-data feed so it pulls each company’s real numbers itself.

The Result

Chris Hohn investment philosophy here: Chris_Hohn_Reasoning_Profile.pdf​

Portfolio analysis here: Portfolio_Hohn_Lens_Analysis.pdf​

I ran a representative mega-cap portfolio through it, the kind a lot of you hold.

The first flag was uncomfortable.

Its biggest single stock was Microsoft, at almost 12 percent. That is the exact name Hohn just cut.

Most of the book sits in the safe zones. One name drifts toward the danger corner.

Here are a few things the AI pulled straight from his own words:

• His real test for a moat is not growth. It is whether a company can price above inflation for years.
• On software his view is blunt. “AI eats software,” and per-seat licenses get re-rated as agents do the work.
• On the Microsoft sale, the trigger was conviction, not price. He thinks AI can rewrite the Office workflow his moat depends on.
• Whole sectors are off-limits to him: banks, insurers, oil. The representative book held all three.

So here’s my read. (1) The portfolio’s biggest position is the one Hohn most wants to avoid. (2) Its three blacklist names are disliked for old reasons, not AI ones.

But the output isn’t really the point.

The real point is that his interview and his letter are public, and almost nobody reads them in full.

AI lets you turn any great investor’s primary sources into a thinking partner, then audit your own book against it in an afternoon.

Excess returns come from a differentiated view versus consensus. This is one way to build one.

I want to note this isn't foolproof. The portfolio here is a representative example, not advice, and the profile is my AI's read of Hohn, not gospel.

Interestingly, he says it best himself. Most investors trust authority too much, so spot-check the quotes before you lean on them.

Try this on an investor you actually follow. Pick their last interview and see how they really think.

Personal

If you want to learn how to wire this whole workflow up yourself, I’m teaching it live in our next Wall Street Prompt cohort.

It’s the same setup I use to go from a raw transcript to a portfolio read in an afternoon.

Details are in the link here:

Hit the reply button if you have any questions!