Ideas worth thinking through
There is a particular kind of intellectual courage that does not get enough attention in investment research: the willingness to revise a carefully constructed view when the evidence genuinely warrants it. Most discussions of investor discipline focus on holding firm against panic or resisting the temptation to chase momentum, and those are real skills worth developing. But the opposite failure is equally dangerous and considerably less discussed. When a researcher clings to a thesis long after the underlying conditions have shifted, not because the logic still holds but because abandoning it feels like an admission of error, the result is not discipline at all — it is a form of self-deception dressed up as conviction. The first step toward healthier practice is simply acknowledging that updating a view and abandoning a view are not the same thing. A genuine update is a structured response to new information that changes the probability of a specific outcome. An abandonment driven by anchoring bias or social pressure is something else entirely, and the two can feel remarkably similar from the inside, which is precisely why a deliberate process matters so much.
The practical challenge is learning to distinguish signal from noise in real time, and this is harder than it sounds because markets generate an almost continuous stream of new data, commentary, and narrative. Not every piece of information that arrives after you form a thesis deserves to alter it. Some developments are genuinely new and materially relevant to the assumptions your thesis rests upon. Others are simply variations on information you already incorporated, repackaged by the news cycle to feel more urgent than they are. A useful discipline is to write down, at the time you form a view, the specific conditions under which you would expect to revise it. What would have to be true — or demonstrably false — for the core logic to break down? This kind of pre-commitment does two things simultaneously. It forces you to articulate the load-bearing assumptions in your thesis rather than leaving them vague, and it gives you a reference point against which to test incoming information rather than evaluating each new development in isolation. When something arrives that genuinely speaks to one of those pre-identified conditions, you have a principled reason to revisit your thinking. When it does not, you have an equally principled reason to set it aside.
Scenario comparison is one of the most underused tools for managing the updating process well. Rather than maintaining a single thesis and defending it against all comers, it is often more productive to hold two or three plausible scenarios simultaneously, each with its own internal logic and its own set of observable indicators. This approach changes the nature of the question you are asking when new information arrives. Instead of asking whether the new information supports or undermines your view, you ask which of your scenarios it makes more or less likely, and by roughly how much. That framing keeps you honest in a way that a single-thesis approach does not, because it forces you to consider the alternative explanations rather than filtering them out. It also makes updating feel less like a defeat and more like a natural part of the research process, because you are not abandoning a position so much as adjusting the relative weight you place on possibilities you were already tracking. The scenarios themselves should be grounded in the structural features of the situation you are analysing — the incentives of key actors, the constraints imposed by the broader environment, the historical patterns that seem most analogous — rather than in short-term price movements or sentiment indicators, which tend to generate far more noise than insight.
Perhaps the most important habit of all is separating the quality of a decision from its eventual outcome. In any domain where uncertainty is irreducible, a well-reasoned view formed on the best available evidence can still prove incorrect, and a poorly reasoned guess can occasionally appear to be vindicated. If you evaluate your updating process solely by whether the revised thesis turned out to be right, you will draw the wrong lessons from experience almost as often as the right ones. What you can evaluate is the process itself: whether you identified the right assumptions to test, whether you sought out information that could challenge your view rather than only information that confirmed it, and whether your revisions were proportionate to the strength of the new evidence rather than to the emotional pressure you felt at the time. Keeping a research log — even a simple one — makes this kind of retrospective review possible in a way that relying on memory alone never can, because memory is notoriously susceptible to the very biases you are trying to counteract. The goal is not to be right every time, which is not a realistic standard in any honest account of how research works, but to build a practice that improves your judgement incrementally over time and remains genuinely responsive to the world as it is rather than as you initially expected it to be.