skills/full-company-analysis/references/appendices/Appendix-M-final-99-point-self-audit.md

<!-- Generated loss-aware reference mirror from God_Level_Public_Company_Financial_Analyst_Job_Guide_V6_99_ALL_SUB70_FIXED.docx. Canonical source remains the bundled DOCX. -->

<!-- Appendix: M | Title: FINAL 99-POINT SELF-AUDIT -->

# APPENDIX M - FINAL 99-POINT SELF-AUDIT

| Dimension | 99-level requirement | Weight |
| --- | --- | --- |
| Breadth | Complete end-to-end research system plus sector adaptation and reusable deliverables | 10 |
| Source discipline | Primary-source-first, versioned, citable, reproducible evidence trail | 10 |
| Accounting/forensics | Three-statement mechanics, complex accounting, non-GAAP, estimates, fraud screens, cash corroboration | 10 |
| Business/industry | Unit economics, pricing, retention, capital intensity, industry structure, competition, disruption | 10 |
| Modeling | Driver-based integrated statements, schedules, scenarios, QA, no hidden plugs | 12 |
| Valuation | DCF, relative, reverse DCF, SOTP, terminal consistency, EV-equity bridge, dilution | 12 |
| Risk/decision process | Liquidity, stress, thesis breaks, pre-mortems, post-mortems, decision journal | 10 |
| Sector specificity | KPIs, traps, valuation, bottlenecks, bear cases, model tabs for major sectors | 10 |
| AI/automation | Provenance, structured extraction, deterministic compute, evals, security, human approval | 8 |
| Teachability | Prerequisites, inputs, procedures, formulas, examples, failure modes, exit criteria, training path | 8 |



> Target: no dimension below 99% of its intended standard. A static manual should not claim absolute 100% coverage because regulations, technologies, products, and market structures evolve. The required regulatory-change monitor and source-verification system are therefore part of the mastery standard.

## Scoring method

- Score each dimension from 0 to 100 using documented evidence. Multiply by the published weight and sum the weighted results.

- 99-level means the dimension is executable by another trained analyst without hidden knowledge, material unsupported claims, unexplained model plugs, or missing validation/falsification logic.

- No averaging around a critical defect: any automatic-fail condition below caps the entire manual or research package below 70 until corrected.

- Reviewers must cite the page/module, defect, economic consequence, remediation and retest result for every score below 99.

## Automatic score caps

- Cap below 70 if a material factual claim cannot be traced to a source, historical financials do not reconcile, the balance sheet does not balance, or the EV-to-equity bridge omits a material claim.

- Cap below 70 if valuation depends on an internally inconsistent terminal growth/ROIC/reinvestment assumption, an unexplained plug, or a forecast that cannot be translated into business-native operating units.

- Cap below 70 if the work uses alternative data or AI-generated extraction without provenance, validation and human review sufficient to reproduce the conclusion.

- Cap below 70 if the downside case ignores liquidity, refinancing, covenant, collateral, dilution, or priority-of-claims issues that could dominate common-equity value.

- Cap below 70 if sector analysis uses generic revenue/margin assumptions while omitting the sector-native driver, accounting trap, valuation claimant level, or binding constraint.

## 99-point remediation loop

1. 1. Identify every dimension or module below 99 and rank by decision impact, not convenience.

1. 2. Convert the weakness into a concrete deliverable: missing source, reconciliation, formula, worked example, stress case, source map, or review control.

1. 3. Repair the underlying analytical procedure, not merely the prose describing it.

1. 4. Re-run the calculation/model/source QA and have an independent reviewer reproduce the result.

1. 5. Re-score using the same rubric. Repeat until no material dimension is below 99 or document the reason the standard is genuinely not applicable.

## Final certification evidence

- Completed scorecard with reviewer name/date and links to evidence.

- List of all defects found below 99 and the exact remediation performed.

- Re-rendered and visually inspected final document with accessibility and table-header checks complete.

- Change log that records content added, removed, consolidated, or corrected during remediation.

# PART XVI - MASTERY LABORATORIES

- Primary sources before narratives.

- Economics before accounting presentation.

- Cash, balance sheet, and dilution before adjusted earnings.

- Drivers and constraints before line-item extrapolation.

- Reinvestment and incremental returns before growth worship.

- Ranges and scenarios before false precision.

- Disconfirming evidence before conviction.

- Liquidity and solvency before multiple compression.

- Process quality before outcome worship.

- Versioned evidence and model changes before memory.

- Automation for speed and consistency, human accountability for judgment.

> The following mastery laboratories convert the operating system into zero-to-expert execution. Use them when a core module identifies what must be done but the analyst needs the mechanics, equations, model architecture, worked examples, edge cases, or reviewer tests required to do it correctly.