> For the complete documentation index, see [llms.txt](https://matterhorn-doc.mometic.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://matterhorn-doc.mometic.com/investigate-the-business/analyst-review.md).

# Analyst review

Read Matterhorn analyst commentary as a structured challenge to the evidence, with a manager synthesis, CAN SLIM context, and visible citation checks.

The analyst review is a way to challenge a candidate before becoming attached to it. Open a company and select **Analysis**. If a review is missing or needs an update, use **Run review**. Follow the stage indicator through the bull, bear and research manager. A review completes when all three roles finish in the same run; citation and fact-check statuses are reported separately.

A user-requested review can proceed while scheduled data work runs. Closing and reopening Analysis reconnects to the saved work. If the waiting screen times out, use **Check review** to check that same job. A timeout in the browser does not by itself mean the background review stopped.

![Company Analysis view with commentary status, three-year and near-term context, and CAN SLIM](https://981865776-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FNOJXLd7UJBniObM136Nu%2Fuploads%2Fgit-blob-a709f75b4f8ae0b3f3a8096291b4a561b76a15b1%2Fanalyst-review.png?alt=media)

## Give each role a job

**Bull Thesis** presents the strongest case supported by the supplied company record. **Bear Thesis** tests the same case for weak economics, risks, contradictions, and inadequate support. **Research Manager Summary** should digest the disagreement: what mechanism matters, whether the measured setup is timely, and what would strengthen or invalidate the case.

The manager is designed to synthesize rather than repeat a list of ratios. Its light guidance concerns research priority and conviction—investigate now, monitor for confirmation, or defer conviction—not automatic trade instructions. Near-term readiness can differ from the three-year business view.

## Read peer context for what it is

Where enough recent observations exist, the manager receives a comparison of standard Matterhorn opportunity scores within the same exact SEC SIC classification. It requires at least five other qualifying companies with scores no more than 90 days old.

This is screening context. It is not a matched competitor analysis, a valuation comparison, or proof of superior business economics. The manager should not invent a peer conclusion when the comparison is unavailable.

## Verify, then judge

Read the citation status beside bull and bear prose. A passed numerical check means the figures matched their cited records under the implemented tests. You still have to assess the reasoning: a weak five-year CAGR does not logically prove that a strong recent recovery is impossible, and one strong quarter does not establish durability.

The manager's concise prose keeps supporting references in the audit record rather than filling the paragraph with citation markers. Qualitative judgment is not marked numerically verified simply for citing real facts. Unresolved synthesis or citation problems remain visible.

## The language model's boundary

Ask Matterhorn and analyst commentary use the configured language-model service. The writer works over retrieved records and deterministic context; it does not set financial scores or ML ranks. Review outputs carry zero scoring weight.

Automatic reviews cover the current top ten companies and refresh after an ingestion cycle adds relevant new evidence. You can request a review for another company; the completed review is saved for reuse. When relevant new evidence arrives, its saved review is marked for an update. A new review does not regenerate every company’s commentary.

> **Pro Tip — Read the bear case before writing your own conclusion.** If it identifies a measurable condition that would break your thesis, put that condition into your follow-up notes. If it only repeats weak logic, go back to the actual facts rather than counting another negative vote.

Related: [CAN SLIM](/investigate-the-business/can-slim.md), [citation checks](/investigate-the-business/evidence-and-citations.md), [Ask Matterhorn](/investigate-the-business/ask-matterhorn.md).