Conviction is a research problem.

Every firm claims conviction, and the claim gets tested at the same moment for all of them: the bad quarter, the down round, the milestone that slipped for the second time, when the position looks worst on paper and the easiest thing to do is quietly stop returning the founder's calls. Some investors hold, and a few add. Others find reasons to be elsewhere. Afterward the difference gets explained as temperament — the patient investor, the nervous one — as if the outcome had been decided by personality.

Temperament is the wrong explanation. What separates the two groups is what they knew on the way in. An investor who did the work — the physics, the competing approaches, the customer's actual constraint, the regulatory path — can tell the difference between a thesis that broke and a timeline that slipped. An investor who bought a narrative cannot, because a narrative offers no way to distinguish signal from noise. When the story is all you own, every wobble looks like the story ending.

Conviction that can't say what would change its mind isn't conviction. It's a position that hasn't been tested yet.

The work happens before the test

The research that produces real conviction is specific and slightly unglamorous, and nearly all of it happens before the check is written. It means writing down what has to be true for the investment to work — not as a slogan, as a list. It means knowing which upcoming milestone confirms or denies each item on that list, and deciding in advance what failure would look like as distinct from delay. Then the bad quarter arrives and the question is no longer the unanswerable “do we still believe?” It is “which assumption moved?” — and that question has an answer someone can go find.

The same file that lets a firm hold also tells it when to stop. Conviction is not loyalty. A thesis with a written falsification condition is a thesis that can actually fail, and an investor who knows what failure looks like can exit a broken position while the narrative investors are still averaging down on faith. Holding correctly and selling correctly are the same skill, exercised in opposite directions.

None of this makes markets kind or timelines honest, and no amount of preparation removes the discomfort of the middle years. What it changes is the source of the steadiness. It relocates conviction from personality to process — from something a firm asserts about itself to something it can build, document, and check. The firms that hold through the middle aren't braver than the ones that don't. They did the reading, and they wrote down what they found.