Of all the structural properties an investment thesis can have, falsifiability is the one that most directly determines whether the thesis can do its job. A thesis that cannot be falsified is a belief. It will survive indefinitely, accommodating every piece of adverse evidence until the loss becomes severe enough to force an exit on emotion rather than logic.

The broader Investment Thesis Checklist covers falsifiability in five questions. This dedicated checklist goes deeper, working through 12 questions organized into four groups: named falsifiers, time frame, evidence threshold, and process. It is designed to be run as a standalone document when falsifiability is the primary gap in an otherwise well-researched thesis.

Why falsifiability deserves its own checklist

Falsifiability is not one quality among many for an investment thesis. It is the quality that determines whether a thesis can function as a decision tool rather than a story.

A thesis with excellent assumption quality, a well-specified time horizon, and clear break conditions but no falsifier is still fundamentally broken as a decision framework. You can monitor the assumptions, but you have no built-in mechanism for deciding when enough assumptions have failed to warrant an exit. You can observe the time horizon approaching, but without a falsifier you will find reasons why the thesis needs more time.

The most common failure mode in long-term positions is not that the investor lacked information. It is that the information was available but there was no pre-committed standard for acting on it. Falsifiability provides that standard. The 12-item checklist in this guide targets the specific failure modes that allow theses to appear falsifiable without actually being so.

This checklist also differs from the broader thesis checklist in one important respect: it is designed to be used not just at entry but at every review date throughout the life of the position. Falsifiability is not a static property. A falsifier written at entry may need to be updated as the business evolves, as new risks emerge, or as the original assumption behind the falsifier is proven stronger or weaker than expected.

The 12-item falsifiability checklist

Answer each question yes or no. Yes earns one point. Count your total at the end of each group and at the end of the full checklist.

Group A: Named falsifiers (4 items)

  1. Is there at least one specific observable falsifier named in the thesis document? The falsifier must be a business or market outcome, not a general description of things going badly.
  2. Is the falsifier different from a price decline or a drop in the stock? Price-based falsifiers fail the test because they reflect market opinion, not business reality.
  3. Would a thoughtful skeptic who disagrees with the thesis accept that this event would break it? If the answer is "they might argue around it," the falsifier is too weak.
  4. Is the falsifier written down in the thesis document itself, not just held in memory? A falsifier that exists only in memory is subject to gradual redefinition under pressure.

Group B: Time frame (3 items)

  1. Is a specific time frame attached to the falsifier? A falsifier without a time frame can always be deferred: "the event hasn't happened yet" is not the same as "the thesis is intact."
  2. Is the time frame proportional to the thesis type? A short-term catalyst thesis needs a short falsifier window. A long-term compounding thesis can have a longer window but should have interim checkpoints rather than one terminal date far in the future.
  3. Was the time frame written down before entry, not at a later date when the original expectation had already proven wrong? A time frame written after the original window has passed is a revision, not a falsifier.

Group C: Evidence threshold (3 items)

  1. Is the evidence threshold specific and, where possible, numeric? "Revenue declines materially" fails. "Revenue declines more than 10% year over year in any reported quarter" passes.
  2. Is the threshold calibrated to something observable and consistently reported: a financial metric, a data release, a regulatory filing, or a public announcement? A threshold tied to an observable data source removes the judgment call from the evaluation.
  3. Would the threshold produce a clear pass/fail signal at the review date, such that two investors looking at the same data would agree on whether the threshold was breached?

Group D: Process (2 items)

  1. Has the falsifier been reviewed by at least one person who disagrees with the thesis, or tested against the strongest available counterargument? The adversarial review catches falsifiers that are technically present but structured to be easily rationalized around.
  2. Has the thesis document been updated to reflect any changes to the business, the competitive environment, or the assumptions since the falsifier was first written? A falsifier written two years ago for conditions that have since changed may no longer be testing what the thesis actually depends on.

Scoring and interpretation

Add up your points across all four groups. The maximum is 12.

12 out of 12: Strong falsifiability

The thesis has a complete falsifiability structure. All four groups pass, including the process items that confirm the falsifier has been adversarially tested and kept current. This is the standard for high-conviction positions where accurate self-assessment of thesis health is most critical.

9-11: Adequate, address the gaps

The thesis is falsifiable at a working level but has identifiable gaps. Review which items scored zero and address them before the next review date. A score of 9 or 10 with all gaps in Group D (process items) is acceptable temporarily: commit to the adversarial review before adding to the position. A score of 9 or 10 with gaps in Groups A or C requires immediate attention.

6-8: Proceed with caution and reduced position size

The thesis has partial falsifiability structure but meaningful gaps. A score in this range with low scores in Group A (named falsifiers) means the core problem remains unsolved: you have not named what would break the thesis. Open a small initial position, commit to a specific date by which the falsifiability structure will be improved, and treat that date as a hard deadline for either improving the score or exiting the position.

Under 6: Revise before entering

Multiple structural gaps exist across the falsifiability framework. This score indicates that the thesis either lacks named falsifiers entirely, lacks a time frame, or has thresholds too vague to be actionable. Do not commit capital until the score reaches at least 9. The gaps in this range are structural and cannot be resolved by additional research: they require explicit commitments about what would break the thesis.

The most common falsifiability failures

Three failure patterns account for the majority of theses that fail this checklist despite appearing falsifiable on the surface.

Pattern 1: The compound falsifier

A compound falsifier requires multiple conditions to be met simultaneously before it triggers. "The thesis breaks if revenue declines AND margins compress AND management credibility is damaged in the same period." Each individual component is observable and specific, but the compound requirement means the falsifier almost never triggers: there will always be at least one component that has not yet crossed the threshold, giving the investor permission to continue holding.

Fix: Decompose the compound falsifier into separate falsifiers for each load-bearing assumption. Any one of them triggering should prompt a review, not all of them triggering simultaneously.

Pattern 2: The retrospective falsifier

A retrospective falsifier is one that is written or revised after the event it was supposed to test has already occurred. "I was watching for revenue to decline more than 15%, and it declined 12%, so the thesis is still intact." If the original written falsifier was "revenue declines materially," the 12% figure is a post-hoc threshold invented to keep the thesis alive. This is among the most common and most consequential falsifiability failures because it occurs at exactly the moment when accurate self-assessment is most needed.

Fix: Write falsifiers with specific numeric thresholds before entry. Review the written falsifier at each review date without modifying it unless you are explicitly reclassifying the thesis, which requires a written explanation of why the original threshold was wrong and what the new one is based on.

Pattern 3: The external falsifier

An external falsifier ties the thesis's break condition to something outside the thesis's core mechanism. "The thesis breaks if there is a major recession." A major recession is a real risk, but it is external to the specific business logic of the thesis. If the thesis is about a company gaining market share through a product advantage, the falsifier should test whether the product advantage is holding, not whether the macroeconomic environment is benign. External falsifiers give the illusion of testability while leaving the actual thesis mechanism unmonitored.

Fix: For each external falsifier, ask what internal mechanism the external event would disrupt. Name that internal mechanism as the falsifier instead. "The company's win rate in competitive sales situations falls below 55% in any two consecutive quarters" tests the product advantage directly, regardless of macroeconomic conditions.

Using this checklist at review dates

The falsifiability checklist is most valuable when used consistently across the life of a position, not just at entry. A thesis that scored 12/12 at entry can drift to 8/12 over two years as the business evolves, as original falsifiers become stale, or as new risks emerge that the original framework did not anticipate.

At each scheduled review date, run the checklist in full and compare it to the score from the previous review. Items that have declined from yes to no require specific attention. Items 11 and 12 (the process items) should be actively maintained between reviews, not just evaluated at review time.

Mark which items have changed status since the last review. An item that moved from yes to no is an action item. An item that moved from no to yes represents a strengthening of the falsifiability structure, which can support a re-evaluation of position size.

One important caution: do not modify the falsifiers themselves at review time without a written explanation. The purpose of a review is to evaluate the thesis against pre-committed standards, not to update the standards to fit the current situation. If a falsifier needs to be changed because the original assumption has been superseded by events, write out explicitly what changed, why the original falsifier no longer applies, and what the new falsifier is testing. An undocumented revision to a falsifier at review time is the retrospective falsifier pattern described above.

Frequently asked questions

What makes an investment thesis genuinely falsifiable?

A thesis is genuinely falsifiable when it names at least one specific observable event that would prove it wrong, that event is a business or market outcome rather than a price movement, the evidence threshold is specific enough to produce an unambiguous signal, and the time frame in which the event must or must not occur is defined before entry. The adversarial test also matters: a thoughtful skeptic who disagrees with the thesis should agree that the named event would falsify it.

How many falsifiers does a thesis need?

At minimum, one falsifier per load-bearing assumption. A thesis with two or three key assumptions should have at least two or three falsifiers, one for each assumption. A thesis with only one falsifier is fragile: if that one falsifier is avoided through rationalization, there is no backup trigger. In practice, two to four well-specified falsifiers is the standard for a thesis with a clear argument structure.

Can a thesis be falsified by qualitative evidence?

Yes, but qualitative falsifiers need to be made as specific as possible to avoid motivated reasoning at the time of evaluation. "Management loses credibility" is too vague. "Management makes a significant capital allocation decision that contradicts the stated strategy in the three-year plan, as announced in a public filing or earnings call" is qualitative but specific enough to produce a clear answer. The question to ask is: would two thoughtful investors reading the falsifier agree on whether the event had occurred?

What should I do if my thesis passes all 12 checklist items but still feels weak?

Trust the feeling and investigate it. A thesis that passes the structural checklist but still feels weak is likely weak in the underlying assumptions rather than in the falsifiability structure. The checklist evaluates whether the thesis is properly structured; it does not evaluate whether the core claim is correct. If the thesis passes 12/12 but feels weak, go back to the load-bearing assumptions and ask whether they have been genuinely tested against evidence or merely asserted.

How often should I review the falsifiability checklist for an existing position?

At each scheduled review date, which should be determined by the thesis type. A short-term catalyst thesis might need monthly review. A three-year value thesis might need quarterly review. A ten-year compounding thesis might need annual review with a brief quarterly check on key leading indicators. The review should answer two questions: have any falsifiers been triggered, and have the falsifiers themselves changed in a way that requires updating the checklist?

References