Prepare an old thesis for a CoTrading review
S02 needs a record of what you previously believed. A company description written today is not an old thesis, and a good recent result cannot establish that an earlier prediction existed.
Freeze the baseline
Copy the original dated note without editing it. Add an input cover sheet with issuer identity, original date, review cutoff and new source list. If the old note is vague, preserve that vagueness as a limitation; do not strengthen it retroactively.
| Original statement | New synthetic evidence | Review state |
|---|---|---|
| Revenue stays in double-digit growth | Comparable growth is 12% | Supported for this period |
| Cash flow / profit stays above 80% | Current ratio is 69.70% | Challenged |
| No major customer losses | No retention data supplied | Unknown |
These fictional DEMO-CONSUMER conditions are user-defined, not universal investment standards. Original note: 2026-08-01. Review cutoff and synthetic evidence release: 2026-10-30. Amounts underlying the ratio are operating cash flow 92 and net profit 132 in the same nine-month window.
Make the handoff explicit
The current 0.1.0 S02 package uses a deterministic Python 3.12 tool, as do S01/S04/S05. Its teaching task needs only the standard library, with no tool network access or third-party Python packages. A text thesis does not remove that runtime requirement. If Python is missing, stop fixed-tool execution and follow the installation guide for isolated setup; a model-only summary is not a completed tool run.
Prompt: “Use cotrading-thesis-check to compare my dated note with new evidence through the cutoff. Do not change the original, invent old reasons or turn unknown into support. Separate proposed revisions from changes in evidence.”
The SEC statement guide is a reference for identifying statement types, not a validation of this hypothesis framework. Attach statement locations and periods to the evidence rather than citing a general guide for the company’s facts.
Review the revision log
A useful output has three artifacts: a condition-by-condition review, a changes file and a next-check list. For the example, the margin of safety is not recalculated and a sell decision is not automated. The only established findings are one supported, one challenged and one unknown condition.
If you accept a revised condition, save a new version with date and reason. Keep “newly added reason” distinct from “old reason supported by new evidence.” Ask for aging and later collection data rather than assuming management’s payment-timing explanation is complete.
When this does not apply
Without a dated baseline, create a new baseline and state that historical drift cannot be assessed. Partial evidence can support a limited review, but cannot establish that all conditions hold. A next-check file is a record, not a scheduled monitor or continuous data service.
FAQ
What if I never wrote my old thesis?
Create a new baseline and disclose that historical comparison is unavailable.
Does a challenged condition mean automatic selling?
No. The review records evidence; the user retains the decision.