Public methodology · Current version v1.1.1

Research Methodology

AegisIQ publishes research produced by a documented scoring framework with versioned methodologies, identified data sources, and human review before publication. This page describes how ratings, valuation scenarios, and AI-assisted narrative are produced, what their boundaries are, and how corrections are handled. It is versioned alongside the methodology itself; historical versions are preserved and each published rating records the methodology version that produced it.

1. Data sources and freshness

Fundamentals, market data, estimates, and transcripts are sourced from licensed market-data providers (currently Financial Modeling Prep) and public filings. Each rating run records the as-of date of its inputs, and published research states the data date it was built from. Where a required input is missing or stale, the affected component is not scored and the outcome rules in section 3 apply — missing data is surfaced, never imputed silently.

2. Rating components and weights

A published research grade is a weighted composite of five components. The weights are locked per methodology version and are, for the current version:

Fundamentals quality — 25%. Balance-sheet health — 25%. Valuation — 20%. Catalyst and momentum — 15%. Risk profile — 15%.

Each component is built from defined sub-signals with published clip ranges. Composite scores map to letter grades (A+ through D) through fixed boundaries; for example, a composite of 95–100 maps to A+ and 90–94 maps to A. Grades describe the structured quality of the research setup for a security. They are not investment advice, and the platform's vocabulary controls prohibit action-oriented language (for example "buy" or "reduce position") in rating output.

3. Research-grade boundaries and insufficient data

A rating is only produced when a minimum number of sub-signals (currently four) are present with usable data. Below that threshold the security is reported as not rated with an insufficient-data reason rather than assigned a grade. Grade boundaries are inclusive integer ranges on a 0–100 composite and are codified in the rating engine; they do not vary by security, sector, or subscriber.

4. Sector substitutions

Some sectors report economics that generic fundamentals do not capture. The methodology applies documented, ratified substitutions in those cases — for example, return-on-equity substitution for banks, custody businesses, asset managers, and utilities (methodology v1.1), and funds-from-operations signals for REITs (v1.1.1). Where a substitution is not yet validated for a cohort, affected securities surface as insufficient-data rather than being scored on inappropriate signals. Every rating row records whether a substitution path was applied.

5. Valuation-scenario construction

AegisIQ publishes hypothetical low, base, and high valuation scenarios constructed from a discounted-cash-flow framework, comparable-company references, and technical reference levels, weighted per the methodology version. Scenarios are illustrations of stated assumptions — they are not price targets, predictions, or guarantees, and they are expressly hypothetical. The assumptions behind each scenario are recorded with the artifact.

6. Methodology versioning and historical preservation

Methodology changes are ratified in written decision records before implementation, bump the methodology version, and apply only from the rating run where they activate. Existing rating rows remain on the version that produced them and are never restated in place. Version history: v1.0 (baseline five-component composite), v1.1 (sector-aware ROE substitution), v1.1.1 (REIT FFO signals).

7. AI narrative limitations

AI-generated narrative summarizes the scored evidence. It can be inaccurate, incomplete, stale, or biased, and model error including hallucination is possible. Narrative output is constrained by vocabulary controls, carries the limitations stated in the Investment Research Disclaimer, and is subject to the human review procedure below before publication.

8. Human approval and correction procedures

The AegisIQ Research Review Officer — a designated role, currently held by H. Wayne Hayes Jr., CEO — holds final authority to approve, reject, or require revision of AI-assisted research before it is published to subscribers. A ticker request may retrieve an already approved, canonical research artifact; if no current approved artifact exists, a draft may be generated, but it enters the review queue and is not published, exported, shared, or displayed as approved research until approved. Each generation and approval decision is recorded with the reviewer, role, decision, timestamp, methodology version, model version, source-manifest hash, artifact hash, and any revision note. Corrections are issued as new superseding records; original records are preserved and retained for at least seven years from creation, and longer under any legal, regulatory, dispute, or preservation hold. The full control description is on the Research Controls page.

9. Publication boundary

AegisIQ research follows a general-publication model. Research is canonical by symbol, as-of date, and methodology version, and the same approved artifact is available to all similarly entitled subscribers. Selecting a ticker selects a publication topic; it does not personalize the analysis. Research does not use a subscriber's holdings, age, income, tax status, investment objectives, risk tolerance, projected cash flow, or personal circumstances, and the platform provides no portfolio-specific recommendation, allocation, suitability determination, individualized alert, trade instruction, discretionary authority, or automatic execution.

This page describes methodology and controls; it is not investment, legal, or compliance advice. See the Investment Research Disclaimer, Terms of Service, and Risk Disclosure.