Building an Investment Thesis

Variant perception: the whole of the idea

The price of a liquid security already contains the consensus view. Buying it because the company is good is a bet that the market has not noticed it is good, which for a widely held large-cap is implausible.

A thesis therefore has to name the difference. What does the market believe, specifically, and what do you believe instead? Both halves need saying. "The market believes margins stay at twelve per cent because that is where they have been for five years; I think the mix shift to services takes them to sixteen by 2029" is a thesis. "This is a high-quality compounder" is a description.

The four questions a thesis must answer

Where a real difference of opinion can come from

There are only a few genuine sources of edge, and being honest about which one you have is more persuasive than claiming all of them.

• Informational: you have done work others have not, usually in a neglected or complex name. Rare in large caps, common in small ones.
• Analytical: the same public facts, read better. Usually a modelling insight, a sum of the parts, or a unit economic that nobody has built.
• Time horizon: you are willing to own something through two bad quarters that a fund reporting monthly is not. The most durable edge available to an individual.
• Behavioural: you are buying what others cannot own for structural reasons, such as an index exclusion or a size constraint.
• Being first is not on this list. Being right earlier than it is obvious usually is.

Making the thesis falsifiable

A thesis that cannot be wrong cannot be tested, and a portfolio manager will find that out in one question. Write down, before you buy, the specific observations that would break it.

Good kill criteria are observable, dated and outside your control. "Subscriber growth below four per cent in either of the next two quarters." "Gross margin fails to reach 41 per cent by the second half." "The regulatory decision goes against them in March." Bad kill criteria are the price falling, which is not information, and the thesis not playing out, which is not a criterion.

The numbers that have to be true

Every thesis reduces to a small number of assumptions doing the work. Find them and state them, because a model with forty inputs conceals which three matter.

Then invert. Rather than forecasting the outcome and deriving a value, take the current price and ask what it implies. If the shares are discounting eight per cent growth forever and you think the honest number is four, you have learned more in five minutes than a week of modelling would have taught you. This reverse approach is also the fastest way to answer the interview question about what is priced in.

Risk, and the difference between risk and volatility

A thesis section on risk that lists market risk, execution risk and competition is a section nobody reads. Useful risk work is specific to this company and ranked by how much it would cost you.

Separate the two categories explicitly. Risks that would break the thesis, which should be your kill criteria. And risks that would hurt the price without breaking the thesis, which are an argument about sizing and timing rather than about owning it at all. Candidates who conflate the two either panic out of good positions or hold bad ones.

Sizing, which is part of the thesis

A view and a position are different things, and the step between them is where judgement shows. Sizing follows from conviction, the shape of the downside, liquidity and how correlated the idea is with what the book already owns.

A useful discipline is to size on the bear case rather than on the base case. If the reasonable downside is a forty per cent fall and you do not want to lose more than two per cent of the fund on any single idea, the position is five per cent, regardless of how good the upside looks. Saying that out loud in an interview signals portfolio thinking that most pitch-focused candidates never display.