name: stock-valuation description: FenokValue v9.6.4 - Korea Discount Reason Code fix. Critical bug fixes. Data contract validation. Global Scouter v2.2.0 integration.

FenokValue Skill Guide (v9.6.4)

엔진: yfinance + Global Scouter v2.2.0 + Damodaran 4-JSON + ValuationOrchestrator
핵심 기능: TWO_STAGE DCF, Reverse/Stress, Reason Codes(134개), 8 L1 Pillars, Masters Panel (4 modes), Data Contract Validation
버전: 9.6.4 (2026-01-29) · Author: El Fenomeno

🔴 CRITICAL: Quick Start First

MANDATORY: Before any analysis task, complete these steps:

Read Reference: Review CLAUDE.md for detailed agent reference
Install Dependencies: pip install -r requirements.txt
Verify Installation: python -c "import yfinance, pandas, numpy; print('OK')"
Test Run: python fenok_value.py report NVDA --format md

⚠️ DO NOT skip these steps. Data fetching requires proper environment setup.

Resources & Folder Structure

stock-valuation/
├── SKILL.md              # This guide (user-facing)
├── CLAUDE.md             # Agent reference (detailed)
├── PROMPT_GUIDE.md       # Prompt examples
├── scripts/
│   ├── fenok_value.py    # CLI entry point (22 subcommands)
│   ├── report_generator.py   # v9.1.0 UnifiedReportGenerator
│   ├── masters_panel.py  # 13 investors, 4 modes
│   ├── valuation_engine.py   # DCF/Reverse/Stress
│   ├── quality_score.py  # Lynch/Piotroski/Altman
│   ├── core/             # 50+ core modules
│   │   ├── valuation_orchestrator.py   # Main orchestrator
│   │   ├── valuation_router.py         # Stock type routing
│   │   ├── stock_data_fetcher.py       # Data fetching
│   │   └── wacc_calculator.py          # WACC engine
│   ├── adapters/         # Data adapters
│   ├── providers/        # DataPack, SEC13F providers
│   └── tests/            # (dev only, excluded from ZIP)
├── masters/              # 13 investor profiles (JSON)
└── references/           # Technical specifications

Dependencies

Package	Version	Purpose
yfinance	0.2.30+	Stock data fetching
pandas	1.5.0+	Data processing
numpy	1.23.0+	Numerical computations
requests	2.28.0+	API calls

1. Installation

cd scripts
pip install -r requirements.txt
python -c "import yfinance, pandas, numpy; print('OK')"

2. 핵심 명령어

목적	명령
Quick 스크리닝	`python fenok_value.py report NVDA --level quick`
Standard 분석	`python fenok_value.py report NVDA --level standard`
Full 보고서 (권장)	`python fenok_value.py report NVDA --level full`
Deep 분석	`python fenok_value.py report NVDA --level deep`
데이터/WACC	`python fetch_data.py NVDA --wacc-breakdown --estimates`
밸류에이션	`python valuation_engine.py NVDA --mode full --format md`
품질 점수	`python quality_score.py NVDA --mode full --view summary`
비교 분석	`python valuation_engine.py NVDA AMD --mode full --compare`
적자 성장주	`python ev_sales.py SNOW --debug`
대가 패널	`python masters_panel.py NVDA --mode debate`
리스크	`python risk_radar.py GME --mode full`

Report Levels (v9.5.0 Hierarchy)

Level	Analyzers	시간	Masters	용도	Mode
`quick`	5개	~10s	❌	Screening (standalone)	orchestrator
`standard`	14개	~2min	❌	Base analysis	orchestrator
`full`	20개	~2min	❌	STANDARD + 6 advanced (기본값)	orchestrator
`deep`	21개	~5min	✅	FULL + Masters	orchestrator

Hierarchy: STANDARD (14) ⊂ FULL (20) ⊂ DEEP (21) | QUICK (5) = standalone

🔥 v9.6.0 Changes:

BUG-009: dividends 섹션 매핑 추가 → DDM 분석 정상화

BUG-010: 성장률 30% Cap 추가 → 고성장주 과대평가 방지

BUG-011: masters_all_modes 임시 제거 (DEEP 22→21)

NEW: Data Contract 검증 (data_validator.py)

🔥 Breaking Change v9.5.0: 계층 구조 재설계

STANDARD: subprocess → orchestrator 모드 변경

Analyzer Names: dcf → conviction_score, relative → relative_valuation, etc.

New Adapters: altman_z, lynch_classifier, relative_valuation (3개 신규)

Hierarchy Enforced: validate_hierarchy() 런타임 검증

⚠️ Deprecated: full 및 analyze 명령은 deprecated 됨.

full → report --level standard 사용

analyze → report --level full 사용

CLI 옵션:

--level: 분석 깊이 (quick/standard/full/deep) - v9.3.0 신규
--view: 출력 레벨 (summary/full/debug)
--mode: 분석 유형 (dcf/reverse/stress/full/debate)
--format: 출력 형식 (text/md/json/html)
--compare: 비교 모드 활성화
--no-masters: Masters Panel 제외 (report 명령)
--no-risk: Risk Radar 제외 (report 명령)

3. v9.5.0 Migration Guide

Breaking Change: Hierarchy Restructure

What Changed:

# Before v9.5.0 (subprocess mode for STANDARD)
STANDARD_ANALYZERS = {fetch, dcf, relative, scenario, risk, ...}  # subprocess names
FULL_ANALYZERS = {conviction_score, reverse_dcf, ...}  # orchestrator names
# STANDARD ⊄ FULL (different naming, different mode)

# After v9.5.0 (all orchestrator mode)
STANDARD_ANALYZERS = {conviction_score, reverse_dcf, relative_valuation, ...}  # 14 analyzers
FULL_ANALYZERS = STANDARD_ANALYZERS | {monte_carlo, sp500_context, ...}  # 20 analyzers
DEEP_ANALYZERS = FULL_ANALYZERS | {masters_panel, masters_all_modes}  # 22 analyzers
# STANDARD ⊂ FULL ⊂ DEEP (strict superset hierarchy)

Why the Change?

계층 무결성: STANDARD가 FULL의 부분집합이 아니었음 → 수정됨
일관성: 모든 레벨이 orchestrator 사용 → parallel 실행 이점
확장성: 새 분석기(altman_z, lynch_classifier, relative_valuation) 추가 용이

Deprecated Analyzer Names (v9.5.0)

Old Name (subprocess)	New Name (orchestrator)
`dcf`	`conviction_score`
`relative`	`relative_valuation`
`scenario`	`scenario_analysis`
`risk`	`risk_checker`
`reverse`	`reverse_dcf`
`stress`	`rate_stress`
`sbc`, `dilution`	`sbc_dilution`
`lynch`	`lynch_classifier`
`masters`	`masters_panel`

Migration Paths

Before v9.5.0	After v9.5.0	Notes
`report --level standard`	No change	CLI 동일, 내부 실행 모드만 변경
`report --level full`	No change	Analyzer count: 18 → 20
`report --level deep`	No change	Analyzer count: 20 → 22

New Analyzers (v9.5.0)

Analyzer	Purpose	Level
`altman_z`	Bankruptcy risk (Z-Score)	QUICK, STANDARD+
`lynch_classifier`	Peter Lynch 6-type classification	STANDARD+
`relative_valuation`	Sector comparison + PER band	STANDARD+

4. 워크플로우 예시

단일 종목 기본 분석

python fenok_value.py analyze NVDA --format md

또는 단계별:

python fetch_data.py NVDA --wacc-breakdown --estimates
python valuation_engine.py NVDA --mode full
python quality_score.py NVDA --mode full

비교 분석

python valuation_engine.py NVDA AMD --mode full --compare
python quality_score.py NVDA AMD --compare

적자 성장주

python ev_sales.py SNOW --debug
python rule_of_40.py SNOW --view summary

4. 데이터 소스 (v9.0.0 Policy)

우선순위	데이터	출처
1순위	EPS/FCF 추정치, PER 밴드	Global Scouter datapack
2순위	주가/재무 (폴백)	yfinance (Lazy init)
3순위	ERP/CRP, 산업 베타	Damodaran JSON (178개국, 96산업)

v9.0.0 핵심: DataPack → yfinance → Damodaran 우선순위. Evidence 추적.

5. 성능 최적화 (v9.2.0)

병렬 처리 설정

구성요소	설정	개선율	근거
데이터 Fetching	`ThreadPoolExecutor(max_workers=5)`	68%	yfinance API rate limit 고려, 5 threads가 안전 마진 확보
Analyzer 실행	`ThreadPoolExecutor(max_workers=6)`	75%	21개 분석기 중 독립적 분석기만 병렬화, 6 threads 최적

벤치마크 결과 (v9.2.0)

구성요소	순차 실행	병렬 실행	개선율
데이터 Fetching	2.20s	0.70s	68%
Analyzer 실행	2.00s	0.50s	75%

벤치마크 스크립트: scripts/benchmark_parallel.py (실행: python benchmark_parallel.py --ticker NVDA)

6. 테스트 (개발 환경 전용)

Note: 테스트는 개발 환경 전용입니다. ZIP 배포본에는 포함되지 않습니다.

7. 버전 히스토리

버전	주요 변경
v9.6.4	Korea Discount Reason Code fix, _generate_market_context_codes() 추가
v9.6.3	Critical fixes: BUG-012 FCF list, BUG-013 Revenue unit, DDM floor
v9.6.0	Root cause fixes: BUG-009 dividends mapping, BUG-010 growth 30% cap, BUG-011 masters removal, Data Contract validation
v9.5.5	Data structure bug fixes: Beta path, EPS estimates, duplicate warnings
v9.5.3	Documentation quality fixes: test stats, ZIP/dev separation
v9.5.2	Claude Desktop bug fixes: P0 info cache, P1 price key resolution
v9.5.1	Bug fixes: P0 race condition, P1 growth fallbacks, P2 logging
v9.5.0	BREAKING: 계층 구조 재설계 (DEC-122), STANDARD ⊂ FULL ⊂ DEEP, 3 new adapters, subprocess 제거
v9.4.0	QUICK 레벨 재설계 (DEC-120), Industry-aligned screening, 3x faster
v9.3.0	`report --level` 통합 (DEC-117), quick/standard/full/deep 레벨
v9.2.1	CLI screening mode added (4 modes complete), 6 masters quantitative fix
v9.2.0	Phase 4 Complete - `insufficient_data` verdict, SKILL.md sections
v9.1.0	Unified Report Generator, Evidence Trail 개선
v9.0.0	Major: Masters Panel 4 modes, DataPack Priority, Global Scouter Full

Last Updated: 2026-01-29

Stock Valuation

Skill Details