There is a meaningful difference between research as an event and research as infrastructure. Most expert programs are built around events: a conviction begins to form, calls get booked, a memo gets written, and the program ends. The intelligence captured in those calls may be excellent, but the moment the program closes, it starts to decay. Market conditions shift, supply chain dynamics evolve, and the expert views that were current in month one no longer reflect reality in month six.
Rolling thesis programs treat expert intelligence differently. Instead of closing the loop at the end of a call sequence, they treat each call as an update to a living document — one that evolves week by week, accumulating evidence, tracking contradictions, and maintaining a running record of how and why conviction has changed. The result is not just a better memo. It is a fundamentally better research process.
This guide explains how to build one — from initial thesis formation through a disciplined six-month cadence to a final investment committee package that is 70–80% complete before anyone drafts a single slide.
“The teams that consistently outperform on primary research don't run more calls. They run better programs. The difference is almost entirely in how they structure continuity.”
What a Rolling Thesis Is (and Isn't)
A rolling thesis is a structured document — maintained by the research team, updated after every expert call or data event — that tracks the evolving state of an investment hypothesis over time. The key word is "updated." Not reviewed at the end of the quarter. Not summarized after the last call. Updated after every call, every time, with a clear record of what changed and why.
It is worth being explicit about what a rolling thesis is not, because many teams believe they are running one when they are not. A rolling thesis is not a list of bullet points from the last call. It is not a static investment memo that gets emailed once and then filed. It is not a summary document that captures what experts said last month without any mechanism for tracking how those views have shifted.
A rolling thesis is a living claim register with version history. It is a confidence-weighted view of each key thesis component — not just an aggregate conviction score, but a disaggregated assessment of which parts of the thesis are well-supported, which remain uncertain, and which have been actively challenged. It is a contradiction log that shows, by name and date, where the thesis has been tested and how the team responded.
The update cadence matters. As a minimum, the rolling thesis should be updated weekly. For programs running four or more calls per month, updates should happen within 24 hours of each call, while the conversation is still fresh and before the next call introduces new complexity. The update need not be extensive — a few sentences explaining what changed and why is sufficient. The discipline of writing something, however brief, forces the analyst to think explicitly about whether the call moved the thesis and in which direction.
Teams that have adopted this approach consistently report that the rolling thesis document becomes the most valuable artifact of the program — not the individual call transcripts, not the analyst notes, but the structured record of how the thesis evolved across all of them.
The Five Components of a Rolling Thesis Document
1. The Core Thesis Statement
The core thesis statement is a single sentence — the specific, testable claim that the expert program is designed to evaluate. It is not a sector view or a general directional bet. It is a precise hypothesis about a specific dynamic that, if true, would make a particular investment compelling. The entire program is organized around testing this statement.
The thesis statement should be written before the first call is booked. If the team cannot articulate the core hypothesis in one sentence at the outset, the program will lack focus. Every expert call, every line in the evidence registry, every entry in the contradiction log should be anchored to this statement. If a call does not bear on the core thesis, it should either be redirected or questioned.
2. The Evidence Registry
The evidence registry is a structured log of every claim made by every expert across every call, tagged to the specific thesis component it supports or challenges. Each entry includes the expert identifier, the date of the call, the claim itself, a confidence assessment, and a notation of whether the claim supports, qualifies, or contradicts the thesis.
The registry serves multiple purposes. It prevents duplication — the team can see at a glance which claims have already been confirmed by multiple sources and which rest on a single call. It also creates accountability for conviction. When an analyst raises the conviction score, the registry provides the evidentiary basis for that decision. Teams that maintain rigorous evidence registries find that their memos write themselves, because the argument has already been assembled claim by claim.
3. The Conviction Score
The conviction score is a numeric assessment of confidence in the core thesis, expressed on a scale of 1 to 10 and updated after each call with explicit written reasoning for any change. The score at program inception reflects the team's prior based on desk research. Every subsequent update is an evidence-based revision.
The key discipline is requiring written rationale for every change. A conviction score that moves from 6 to 7 after a call should be accompanied by a sentence explaining why: which claim moved it, how confident the expert was, and whether the claim was corroborated by prior calls. This prevents the score from drifting upward or downward through accumulated optimism or pessimism without evidentiary justification.
4. The Contradiction Log
The contradiction log is a dedicated record of every claim that challenges the core thesis, with the source, confidence level, and current resolution status. Most teams resist building this — no analyst enjoys cataloguing evidence against their own thesis. But the contradiction log is one of the most valuable outputs of the entire program.
A well-maintained contradiction log shows the investment committee that the team has encountered and grappled with the hard cases — not just assembled corroborating evidence. It also forces resolution: each entry should have a status (open, explained, refuted, incorporated) that indicates how the team has addressed the challenge to the thesis. Unresolved contradictions at month six are the most important questions for the final round of calls.
5. The Open Questions List
The open questions list is a prioritized register of everything the program has not yet answered. It begins at program inception, with the team's initial unknowns, and is updated throughout — both adding new questions that emerge from calls and removing questions that have been satisfactorily resolved.
Priority ranking matters. Not every open question carries equal weight. The team should distinguish between questions that are thesis-critical (unanswered, they undermine the investment case) and questions that are interesting but not load-bearing. Call planning in months four and five should be driven almost entirely by the highest-priority open questions. At program close, the remaining open questions become the foundation for any follow-up research program.
A Month-by-Month Cadence for a 6-Month Program
The following cadence represents a disciplined approach to a six-month expert program. The call volumes are illustrative — the exact numbers will vary by sector complexity and expert availability. The structural logic, however, applies broadly.
Month 1: Thesis Formation. Run three to four calls with broad-scope experts — those with wide sector visibility rather than narrow functional expertise. The goal is not to test the thesis yet, but to establish a baseline conviction score and build the initial claim register. By the end of month one, the team should have a populated evidence registry, an initial conviction score with written rationale, and a first draft of the open questions list.
Month 2: Pressure Testing. Run four to five calls explicitly targeted at thesis weaknesses. The call guide should be designed around the most challenging open questions and the highest-priority potential contradictions. The first entries to the contradiction log are expected in this month — and their absence should be treated as a warning sign, not a positive signal. If no one is pushing back on the thesis, the team is not talking to the right experts.
Month 3: Deep Dive. Run four to five calls focused on the most uncertain thesis components — the elements that the evidence registry shows are either weakly supported or actively contested. The conviction score should be updated meaningfully by the end of this month. If the score has barely moved since month one, either the calls are not being directed at the right questions, or the thesis components are not being disaggregated with sufficient precision.
Month 4: Lateral Validation. Run two to three calls with adjacent experts — suppliers, customers, former competitors, or others who interact with the space but from a different vantage point. These calls are designed to surface blind spots that sector insiders are too close to see. They also help stress-test claims that have accumulated strong corroboration from within the sector. Confirmation from a different angle carries different evidentiary weight than another call with a former operator.
Month 5: Synthesis Draft. Reduce call frequency significantly. The research team's primary effort this month is drafting the investment memo from the rolling thesis document — not conducting new calls. The evidence registry, conviction score history, and contradiction log should provide the structural backbone of the memo. Any gap that becomes apparent during drafting identifies the specific questions that require the final round of calls.
Month 6: Committee Preparation. Run one to two final calls only — on the outstanding questions identified during month five drafting. The goal is to close the specific evidential gaps, not to generate new material. By the end of month six, the rolling thesis document should be complete, the memo should be substantially drafted, and the investment committee package should require only synthesis and formatting.
“The point of a rolling thesis isn't to have more calls. It's to never waste a call on something you already know.”
— Research operations lead, growth equity firmManaging Thesis Drift
Thesis drift is one of the most common — and least discussed — failure modes in extended research programs. It occurs when the investment thesis changes without anyone explicitly deciding to change it. Over the course of a six-month program, thesis drift can shift the team's hypothesis so far from its original formulation that the evidence collected in months one and two no longer maps cleanly onto what the team is now trying to prove.
The causes are typically gradual and informal. Each call slightly shifts the framing of a key claim. An analyst softens a conviction statement in the writeup to be safe. A strong expert view on a peripheral point gets incorporated into the core thesis without explicit deliberation. The conviction score is adjusted in small increments, each individually defensible, but collectively representing a significant departure from the original hypothesis.
“We discovered that our thesis had drifted significantly from the original hypothesis by month four. Nobody had made a deliberate decision to change it. It just eroded across a dozen small conversations.”
Prevention requires structural safeguards. The most effective is requiring written rationale for any change to the conviction score — a discipline that makes implicit drift visible, because the analyst must articulate the specific evidence that justifies the movement. Version control of the thesis document serves the same function: if the core thesis statement has changed between version three and version seven, someone needs to explain why.
A best-practice safeguard is requiring portfolio manager sign-off on any change to the core thesis statement. The thesis statement is not an analyst document — it is a joint commitment about what the program is testing. Changing it is a substantive decision that warrants explicit deliberation. Teams that require this sign-off find that thesis drift is almost entirely eliminated, because the friction of the approval process forces clarity about whether the team is updating evidence or changing the hypothesis.
Scheduled thesis health checks at months two and four provide an additional control. These are brief sessions — thirty to forty-five minutes — where the full research team reviews the rolling thesis document against the original version, explicitly asks whether the thesis has drifted, and if so, determines whether the drift reflects genuine learning or accumulated informality. Most teams that run these sessions find at least one instance of unintended drift in each program.
What a Good Rolling Thesis Looks Like at Month Six
A well-run six-month rolling thesis program produces a specific set of artifacts. The evidence registry should contain between 50 and 120 attributed claims across all calls — the range depends on program intensity and thesis complexity, not on how many words each analyst wrote. Each claim should be tagged to a thesis component, assessed for confidence, and marked as supporting, qualifying, or contradicting the thesis.
The conviction score history should tell a legible story. An experienced reader should be able to look at the score at month one, month three, and month six and understand the arc of the research process — where confidence grew, where it plateaued, and whether any calls materially changed the direction. A conviction score history that moves in a straight line upward, with no inflection points, is almost certainly not reflecting the actual complexity of the research.
The contradiction log should contain three to seven documented contradictions with resolution status. Fewer than three, and the team has not been stress-testing the thesis adequately. More than seven unresolved contradictions, and the thesis itself may require fundamental reconsideration before it is ready for committee.
The final open-questions list — the questions the program did not answer — should be treated as the starting point for any follow-up research rather than as a sign of failure. A program that closes with three clearly articulated, well-prioritized open questions has done its job. Those questions become the thesis for the next program, and the intelligence infrastructure built over six months carries forward into the next research cycle.
At this stage, the investment memo should be 70–80% complete from the rolling thesis document alone. The structure, the key claims, the evidence base, and the documented counterarguments are all present. What remains is synthesis, framing for the committee audience, and formatting. A team that has maintained the rolling thesis rigorously should be able to complete the memo in a fraction of the time it would take to draft from scratch.
Schedule a thesis health check at months two and four. These thirty-to-forty-five minute sessions with the full research team are the most effective safeguard against drift and the most reliable way to ensure the program is still testing what it set out to test.
Building Research Infrastructure That Compounds
Rolling thesis programs require more discipline at the start — establishing the document structure, agreeing on the update protocol, and building the team's habit of treating each call as an update rather than an event. That upfront investment is real. But the payoff comes in compounding returns: each subsequent program benefits from the infrastructure built in the first, and from the team's increasing fluency with the methodology.
The research team that builds this infrastructure once carries it forward indefinitely. The evidence registry format, the conviction scoring protocol, the contradiction log template — these are reusable. The discipline becomes cultural. Teams that have embedded rolling thesis methodology report that they can no longer imagine running a program without it, not because the methodology is complex, but because it makes the value of each call so much more legible.
A well-run rolling thesis is also the best available defense against information decay. Expert views age. Market conditions shift. Regulatory landscapes evolve. In a program built around discrete events, the intelligence from month one is already partially stale by month six. In a rolling thesis program, the intelligence never stops being updated. The thesis document at month six reflects not just the cumulative evidence of the program, but the team's active engagement with how that evidence has evolved — and that is a meaningfully different and more durable foundation for an investment decision.