Why Your Investment Thesis Should Be a Living Document

Your investment thesis should be a living document because the reason you bought something can quietly stop being true, and you will not notice unless you wrote the reason down in a form you can check. A thesis is not a one-time decision made at the moment of purchase. It is an object you maintain: a small set of claims, the drivers behind them, and the specific guideposts you will grade quarter by quarter.

Most investors treat the thesis as a feeling that hardens at the point of buying and then never gets touched again. The price moves, the story in their head updates to fit the price, and the original argument is never revisited. We are concerned here with the opposite habit: keeping the thesis alive as a written object, and knowing exactly what should make you re-open it.

The thesis is an object, not a moment

When people say “my thesis on this company,” they usually mean a general good feeling: strong brand, good management, growing market. That is not a thesis you can maintain. It cannot be graded, so it cannot be wrong, so it never changes.

A maintainable thesis is written down as a handful of specific, falsifiable claims. Falsifiable means each claim could, in principle, turn out false, and you would know it when it did. “The company will keep growing” is not falsifiable in any useful way. “Volume growth stays in high single digits and margins hold in the high teens over the next two years” is. You can check it. You can be wrong about it. That is the point.

The investment memo is where this object is born. A good memo already commits to a claim, states a variant perception (the specific way your view differs from consensus), and lists what would prove the thesis wrong. The problem is that the memo is usually written once and then filed. Treating the thesis as a living document means the memo is the first version of a record you keep updating, not a brochure you write to get the trade approved and then never open again.

Three things belong in that record, and each one is what you will grade later:

• The falsifiable claims. The one or two sentences that say what you believe and why the market disagrees. If either changes, the position changes.
• The drivers. The small number of variables that actually move the outcome. For most businesses this is three to five things, not thirty. Volume, price, margin, a key input cost, maybe a capex cycle. Name them explicitly.
• The guideposts. Dated, specific expectations you can check reality against. Often these come from management guidance or from your own forecast. They are the tripwires.

A worked example of a guidepost

Guideposts are the most concrete part of a living thesis, so they are worth showing rather than describing. Management guidance is a natural source because it comes with a number and a date attached.

Take Asian Paints. On its forward outlook, management guided to high single-digit volume growth, in a band of roughly 8-10%, and pointed to an EBITDA margin band of 18-20%, speaking of maintaining its margin guidance. (EBITDA margin is operating profit before interest, tax, depreciation, and amortisation, as a share of revenue.)

If you own or follow this company, those two bands are guideposts you write down verbatim, with the date and the source. Then each quarter you do one simple thing: check whether reality is tracking toward the band or drifting away from it. Not whether the stock went up. Whether the actual reported volume growth and margin are moving toward 8-10% and 18-20%, or away.

You are watching two things at once. First, the numbers. Is volume growth landing inside the band, above it, or below it as quarters print? Is margin holding in the high teens or slipping? Second, and just as important, the language. Does management still say “maintaining our guidance” with the same confidence, or does the tone shift from confident to cautious, with new hedges appearing, the band getting wider, or the timeline getting pushed out? A guidepost that is quietly walked back tells you as much as one that is missed outright. Our companion piece on how management guidance works goes deeper on reading that language.

None of this is a view on whether the company is worth owning. It is a method: record the promise, then grade reality against it, calmly, on a schedule. The number is neutral. The discipline is the whole point.

What should trigger a re-underwrite or an exit

A living thesis needs rules for when to re-open it. Otherwise you either never revisit it, or you fiddle with it constantly in response to the share price, which is worse. The useful triggers are events, not price moves. Here are the main ones.

A driver breaks. You named three to five variables that drive the outcome. If one of them behaves in a way your thesis did not allow for, the thesis is now running on a broken assumption. This is not automatically a sell. It is a mandatory re-underwrite: you re-open the document and ask whether the position still makes sense once you swap the broken assumption for the real one.

Guidance is walked back. When management retracts or softens a number they previously committed to, one of your guideposts just failed. That does not mean the thesis is dead, but it means the forward claim you were relying on is weaker than it was, and you owe yourself a fresh look at whether the rest of the argument stands without it.

Reality drifts from a guidepost. Sometimes nothing breaks dramatically. The numbers just quietly drift, quarter after quarter, away from the band you recorded. Drift is easy to miss precisely because no single quarter is alarming. This is why you write the guidepost down: so a slow drift becomes visible as a pattern instead of hiding inside your updated-to-fit-the-price memory.

Your variant perception becomes consensus. This is the trigger almost nobody watches, and it is one of the most important. Your edge on a position was that you saw something the market did not. If everyone now agrees with you, that edge is gone, even if you were right. The reason to hold has changed from “the market is wrong about this” to “the market already agrees, and it is priced in.” That is a genuine reason to re-underwrite, not because anything went badly, but because the game you were playing is over.

Put simply:

A thesis dies not only when it is proven wrong, but also when it is proven so obviously right that everyone else believes it too.

Why most investors never revisit

If this is so sensible, why is it rare? A few honest reasons.

The first is that the original decision is emotionally settled. Buying is a commitment, and re-opening the argument feels like inviting doubt into a decision you have already made peace with. It is more comfortable to let the thesis calcify.

The second is that there is usually nothing to revisit. If the thesis was never written down as specific claims and guideposts, there is no record precise enough to grade. A vague good feeling cannot be checked against reality, so reality never gets to correct it. The fix is upstream: write the falsifiable version at the start, in the memo, so there is something concrete to return to.

The third is that revisiting is work, and it competes with the far more exciting work of finding the next idea. New positions feel like progress. Maintaining old ones feels like admin. So the maintenance quietly does not happen, and the book fills with theses that no longer describe why the positions are held.

There is a mechanical trap hiding here too. When you finally do go back to check “was I right,” you are usually looking at data as it reads today, not as it read on the day you decided. Companies restate history: they redraw segments, fold in acquisitions, strip out discontinued operations. So the past you are grading yourself against may not be the past you actually faced. Grading a thesis honestly means remembering what was knowable when the call was made, which is the same problem we cover in what lookahead bias is.

Keeping the document alive in practice

The mechanics are lighter than they sound. You do not need a system. You need a place where the thesis lives, and a rhythm for touching it.

Write the thesis as claims and guideposts in the memo. Note the drivers and the guideposts explicitly, with dates. Then set a cadence, quarterly is natural because that is when companies report, and on each pass do three things: mark each guidepost as tracking, drifting, or broken; note any change in management’s language; and write one line on whether the variant perception still holds. That one line is the most valuable sentence in the whole exercise, because it forces you to say, out loud, whether you still know something the market does not.

Between scheduled passes, let the triggers pull you back in. A driver breaking, guidance being pulled, a guidepost clearly missed: any of these means you open the document now, not next quarter.

This is the position-level version of a habit that also matters at the level of a whole book. Keeping every thesis current across dozens of holdings is the discipline behind monitoring a portfolio of holdings, and the reason we think continuous research is a real competitive edge rather than a nicety. But it starts here, with one thesis, treated as an object you maintain rather than a decision you made once and walked away from.

The investors who compound tend not to be the ones with the best entry stories. They are the ones who kept grading themselves honestly, quarter after quarter, and closed the gap between what they said would happen and what actually did.