Version an investment thesis without rewriting history
Version an investment thesis by preserving the prior claim and adding a dated explanation of what changed. Editing the old note until it matches the latest evidence makes the research history look cleaner while destroying its usefulness. A version record should show both the information change and the judgment change.
Separate three kinds of revision
A factual correction repairs an error in a source or your transcription. A measurement revision changes the population, definition or accounting basis. A thesis revision changes the explanation or its necessary conditions. These can happen together, but each needs its own description.
IAS 1 describes comparative information and circumstances involving restatement or reclassification. That is a reminder to check whether a new series is comparable with your old one. Your research version history is separate from the issuer’s formal accounting obligations.
Hypothetical revision log
Assume the original thesis says recurring service revenue is growing among existing customers. A later filing separates a newly acquired service business and revises prior-period presentation. The previously copied total remains traceable, but its interpretation changes.
| Version | Change | Reason | Effect on conclusion |
|---|---|---|---|
| 1.0 | Existing-customer growth expected | Total service revenue used as proxy | Tentative support |
| 1.1 | Correct copied unit from thousands to millions | Transcription correction | No change in growth percentage; scale corrected |
| 2.0 | Separate acquired and existing services | New scope disclosure | Original proxy insufficient |
| 2.1 | Require comparable customer cohort | Measurement improvement | Conclusion unresolved until evidence arrives |
Keep the exact date and source for every row. Do not label version 2.0 a mere wording update if it removes the evidence supporting the original conclusion. Conversely, correcting units does not necessarily reverse a growth-rate argument; explain the actual consequence.
Preserve original and comparable series
Store the originally observed value, the latest restated value and the reason for the difference. For analysis across periods, use a consistent series where available. For evaluating your historical decision, use information actually available then. Both views are legitimate, but combining them into one number conceals hindsight.
A minimal change note includes the old claim, new claim, triggering evidence, affected assumptions, evidence publication date and next test. Add whether the old invalidation condition had already triggered before you revised it. This is especially important when widening a time window or replacing a weak metric.
Verify that links point to the correct document version. An issuer website may replace a file under the same address. Document title, filing date and the relevant table location help identify which version you read. Keep retained excerpts proportionate and respect source rights.
This method is useful for evolving research, but it cannot recover an unrecorded historical judgment reliably. If the original version is missing, label any reconstruction as retrospective rather than presenting it as a contemporary note.
A further check on the example
Use version identifiers in downstream notes. If a colleague cites the old service-growth conclusion, the citation should show which claim version it refers to and where a later revision can be found. Otherwise a corrected thesis can coexist with stale summaries that keep spreading the original inference. Review the most important dependent tables after a measurement change; updating the prose alone is insufficient when calculations still combine acquired and existing customers.
FAQ
Does every grammar correction need a new major version?
No. Preserve material changes to evidence, measurement and reasoning. Minor editorial changes can share a small revision note without obscuring meaningful updates.
Can a new thesis reuse the old title?
Yes, if the version history clearly identifies the change in mechanism. A substantially different explanation may be easier to review as a new linked record.