Kill Criteria in Investing
The discipline that compounds is not the trade you put on — it's the exit you wrote down before you needed it.
What are kill criteria in investing?
A kill criterion is a pre-written condition that automatically forces you to sell a position. It is decided at the moment of underwriting — when emotion is lowest and reasoning is clearest — and binds your future self to act when one of a small set of named conditions is met. The point is to remove discretion from the moment of loss, when discretion is most expensive.
Most retail investors who think they have rules actually have moods. They tell themselves at the bar that they would sell if the thesis broke, and then they sit through three years of the thesis breaking because they fell in love with the ticker. The kill criterion is the device that prevents the love affair from costing you the rent. It is written down — in a journal, in your notes app, in the thesis field on your watchlist — and it names a specific condition. Not "the stock falls a lot." Not "the news gets bad." A specific, falsifiable condition that you could read aloud to a stranger and they would know if it had happened.
The pattern is borrowed from poker, where the best players decide their exit before they see the next card. The decision is cleaner in advance because the chips aren't on the table yet. Apply the same rule to a stock: decide what would make you sell before you own it, while the only emotion in the room is curiosity.
How are kill criteria different from a stop-loss?
A stop-loss is a price rule (sell at −15%). A kill criterion is a thesis rule (sell if the moat is gone, the management lied, or the kill condition you wrote at entry has tripped). Stop-losses fire on noise; kill criteria fire on the underlying business breaking. Both can co-exist, but the discipline that actually compounds is the second one — Buffett doesn't trade out of Apple if it falls 15%; he sells if the iPhone moat erodes.
The deeper reason stop-losses underperform thesis rules is that price is a function of the market, and the market is a voting machine in the short run. Stops fire on the vote. Kill criteria fire on the weighing. Across a fifteen-year horizon — which is the only horizon that matters for a compounder — the weight is the thing.
Use both if you must, but never let the stop overrule the kill. A position that has hit a price stop but whose thesis is intact is a position you should be adding to, not closing. Conversely, a position whose thesis has broken but whose price has held up is the most dangerous setup on a screen — the market has not yet caught up to the business, and you have a closing window to leave before it does.
What are the six most useful kill criteria?
Thesis broken (the specific claim you made at entry is now demonstrably false), moat erosion (a competitor closed the gap that justified the premium), management failure (capital allocation is destroying value or trust is gone), structural industry decline (the entire pond is drying up), valuation extreme (the market has already paid you for ten years of growth), and personal discipline violation (you concentrated past the size you said you would underwrite).
You don't need all six. Most positions are well-served by three: the thesis claim itself, a moat condition, and a leverage or governance boundary. The point of the list is not to pick the most criteria — it's to pick the right ones for what you actually believe about the business. A SaaS platform you bought for net retention has a kill criterion built into the metric; a cyclical you bought at trough margins has a different one entirely.
The discipline is to keep the list short enough to remember and specific enough to be objective. "Management blew it" is not a kill criterion. "Capital allocation: net dilution above 5% for two consecutive years without a clear ROI on the issued equity" is.
How do I tell a broken thesis from a temporary setback?
Re-read what you wrote at entry. A temporary setback is a price move with no change to the written claim — earnings missed by 4¢, a class-action got filed, the macro is hostile. A broken thesis is a change to the underwriting itself: the moat you cited has new competitors at parity, the management you trusted is gone, the unit economics you modelled don't hold. If you can't read your own thesis and find the answer, you didn't write the thesis tightly enough.
This is where the work you did at entry pays off. The thesis you wrote — "Apple's moat is the iOS ecosystem and the services attach rate, therefore the right gross margin floor is 38% and the right multiple is 22×" — gives you a sentence you can test. Three years later, when the price is down 35% and you don't know whether to add or fold, you read the sentence and ask: is the ecosystem still real? Is the services attach rate still climbing? Is the margin still above 38%? If yes, the thesis is intact and the kill criterion has not tripped. If no, it has.
The investors who survive forty-year careers all write things down. Buffett's letters are public; Klarman's memos are quiet but exist; Druckenmiller's edge is, by his own description, the ability to re-read what he said about a position and admit when it's no longer true.
When should kill criteria NOT fire?
When the thesis is intact and only the price has moved. The single most common error in self-managed portfolios is treating a 30% drawdown in a quality compounder as a thesis failure. The market voting machine is loud in the short run and irrelevant in the long run; kill criteria attach to the weighing machine, not the voting machine.
A useful test: imagine the position fell another 40% from where it sits today. If the answer is "I'd buy more, because the thesis I wrote is still real and the market is just wrong," the kill criterion has not tripped — the only thing that has tripped is your stomach. That's a problem for the position sizing module, not the kill criterion module.
The most expensive mistake retail value investors make is to fire a kill criterion that was a stop-loss in disguise — to sell a Microsoft or a Costco at a 30% drawdown because the price hurts, then watch it recover and compound for the next two decades without them. The kill criterion is there to stop you killing live businesses, not to give you an excuse.
Do kill criteria belong on every position?
Yes. Even concentrated positions in companies you'd 'hold forever' need a written exit case — because 'forever' assumes the moat, management, and unit economics that justified entry. If those break, forever ends. Writing the kill criteria at entry is also a thesis-testing exercise: if you can't articulate what would make you sell, you haven't underwritten clearly enough to buy.
The hidden gift of kill criteria is what they do to entry. Forcing yourself to write the exit before the entry surfaces theses you couldn't actually defend. If, asked to write a kill criterion, you find yourself writing "if it goes down a lot," you don't have a thesis — you have a hunch. Hunches are fine; they just shouldn't be funded with serious capital.
This is one of the reasons the company scoring methodology we use scores risk and management explicitly: the kill criteria a real thesis requires fall out of those scores. The score is a forcing function; the kill criterion is what the forcing function produces.
What this means for your watchlist
For every position you currently hold, write one sentence: "I will sell this if ____." If you can't fill the blank, you don't understand the position well enough to own it at size. Reduce accordingly, or take a week to do the work and re-underwrite. The moment the kill sentence is written and stored where future-you can find it, you have done more for your real returns than any chart pattern or earnings preview will ever do.