The One-Page Thesis Tracker: Pre-Committing to What Changes Your Mind

Most investors can tell you why they bought a stock. Far fewer can tell you, in writing, what would make them sell it. That gap is where most of the damage gets done — not in the buy, but in the slow drift of holding something whose original reason for being in the portfolio quietly evaporated two years ago.

A one-page thesis tracker is the cheapest fix for this. One page per holding. Written when you buy, reviewed every quarter, and — critically — written in a way that pre-commits you to action. Below is the structure I use, the traps to avoid, and how to know when a broken thesis is actually broken versus just noisy.

What goes on a one-page thesis tracker

Keep it boring. The page should fit on a screen without scrolling. Six fields, in this order:

1. The one-sentence thesis. Not a paragraph. One sentence. "Costco compounds membership fees and gains share as a deflationary retailer." "ServiceNow is the system of record for enterprise workflow and pricing power holds at 20%+ deal expansion." If you can't write it in a sentence, you don't own it — you're renting it.

2. The 2-3 measurable drivers. What actually has to be true for the thesis to play out? Revenue growth above X%? Gross margin holding above Y%? Subscriber net adds positive? Be specific. "Strong execution" is not a driver.

3. Valuation context at entry. What multiple did you pay, and against what assumption? Writing this down stops you from later pretending you bought a growth story when you actually bought a re-rating bet.

4. The kill criteria. Three to five specific, observable conditions that would break the thesis. This is the part everyone skips and the part that matters most. More on this below.

5. The noise list. Things that will move the stock but should not change your mind. A guide-down on Q3 due to FX. A short-seller report on a side business. A competitor's product launch. Naming these in advance is how you keep your hands off the sell button on a bad Tuesday.

6. The next checkpoint. A date and an event — usually the next earnings print, sometimes a product cycle or regulatory decision.

How to write kill criteria that actually work

Kill criteria are where thesis trackers fail. People write things like "if fundamentals deteriorate" or "if management loses credibility." Useless. By the time you agree those statements are true, the stock is down 40%.

Good kill criteria share three properties:

• Observable from public data. You should not need insider information to know whether the criterion has been hit. "Two consecutive quarters of negative organic revenue growth" works. "If they lose their innovation edge" does not.
• Pre-committed thresholds. Not "margins decline" but "gross margin below 38% for two quarters." The number forces you to do the work upfront rather than rationalize in the moment.
• Time-bound. A single bad quarter rarely kills a thesis. Two or three in a row often does. Build the patience window into the rule.

A worked example. If your thesis on a software name is "durable 20% growth at expanding margins," reasonable kill criteria might be: (a) net revenue retention below 110% for two consecutive quarters, (b) operating margin contracting year-over-year for three quarters despite revenue growth, (c) CFO or CEO departure without a credible internal successor. Three specific, observable, time-bound conditions.

When to actually act — distinguishing broken from noisy

Here is the hard part. Kill criteria are necessary but not sufficient. Stocks throw off false signals constantly. The discipline is to treat a tripped criterion as a forcing function for a review, not an automatic sell.

The review asks three questions:

1. Is the criterion actually hit, or am I rounding? If you said "two consecutive quarters" and you have one, you do not have two. Do not negotiate with yourself.
2. Is the cause structural or cyclical? A cyclical miss in a cyclical business is noise. A structural miss — share loss to a new entrant, a pricing reset, a regulatory hit to the unit economics — is signal.
3. Would I buy it here, today, fresh? If the answer is no, the position is being held by inertia, not conviction.

If two of three questions point to "sell," trim or exit. If one does, write down what you're watching for at the next checkpoint and move on. The point is not mechanical selling — it is forcing a deliberate decision instead of a default.

Review cadence and what to log

Quarterly is the right cadence for most holdings. Tie reviews to earnings; the data you need is already in front of you. Spend fifteen minutes per name. Update the drivers with actuals, note whether any kill criteria have moved closer or further away, and either re-confirm the thesis or downgrade it.

Three status levels are enough: on track, watching (one driver weakening, no criteria hit), breaking (criterion hit or close). Once a position goes to breaking, it gets a 30-day decision deadline. No indefinite watchlists.

Keep the prior versions. The most valuable artifact you will build is a record of how your own thinking evolved — including the times you were wrong about what would matter.

What to watch next

• Write trackers for your three largest holdings this week. Not all of them — just the ones where being wrong hurts most.
• Audit your current kill criteria for the "observable and time-bound" test. Rewrite any that fail.
• At your next earnings review, force a status rating (on track / watching / breaking) for every name. No abstentions.
• Track one metric you didn't expect to matter. The drivers you wrote down at entry are often not the drivers that end up deciding the outcome. Notice when a new one starts mattering — that's usually the early signal.