本文へ移動
cccskills
無料GitHub で公開

valuation-model

Valuation methodology — absolute valuation with DCF / DDM / SOTP, relative valuation with PE-Band / PB-ROE / EV-EBITDA, sensitivity analysis, and valuation-trap detection.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md8.9 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Valuation Methodology

Overview

Systematic corporate valuation framework covering absolute valuation (DCF / DDM / SOTP) and relative valuation (PE / PB / EV-EBITDA), including sensitivity-analysis methods and a checklist for identifying valuation traps.

Absolute Valuation Methods

1. DCF (Discounted Cash Flow)

Core formulas:

Enterprise value = Σ FCF_t / (1+WACC)^t + TV / (1+WACC)^n
Equity value = enterprise value - net debt
Per-share value = equity value / total shares outstanding

Detailed steps:

Step 1: Forecast free cash flow (usually 5 years)

FCFF = EBIT × (1-tax rate) + depreciation & amortization - capex - increase in working capital

Simplified version:
FCFF ≈ operating cash flow - capex
YearRevenue (100m RMB)EBIT (100m RMB)FCFF (100m RMB)Growth
2026E1202418+15%
2027E1382821+15%
2028E1553124+12%
2029E1703426+10%
2030E1823628+7%

Step 2: Calculate WACC

WACC = E/(D+E) × Ke + D/(D+E) × Kd × (1-T)

Ke (cost of equity) = Rf + β × (Rm - Rf)
  - Rf: 10-year government bond yield (about 2.5% for China A-shares)
  - β: industry average or company beta (1.0-1.5)
  - Rm-Rf: equity risk premium (about 5-7% for China A-shares)

Kd: cost of debt (loan rate, about 4-5%)
T: income tax rate (25%)

Reference WACC ranges for China A-shares:

IndustryWACC RangeReference β
Consumer8-10%0.8-1.0
Technology10-13%1.2-1.5
Financials7-9%1.0-1.2
Cyclicals9-12%1.0-1.4
Utilities6-8%0.5-0.8

Step 3: Terminal Value

Perpetual-growth method: TV = FCF_n × (1+g) / (WACC - g)
  - g: perpetual growth rate (usually 2-3%, should not exceed GDP growth)

Exit-multiple method: TV = EBITDA_n × EV/EBITDA multiple
  - Reference the industry average or historical median

Step 4: Sensitivity Analysis

### DCF Sensitivity Analysis (per-share value, RMB)

| WACC \ g | 2.0% | 2.5% | 3.0% |
|----------|------|------|------|
| 9.0% | 32.5 | 35.8 | 40.2 |
| 9.5% | 28.3 | 30.8 | 34.0 |
| 10.0% | 24.8 | 26.7 | 29.1 |
| 10.5% | 22.0 | 23.5 | 25.3 |
| 11.0% | 19.6 | 20.8 | 22.2 |

2. DDM (Dividend Discount Model)

Applicable to: high-dividend stocks (banks, utilities, mature consumer companies).

Two-stage DDM:
P = Σ D_t / (1+Ke)^t + D_n × (1+g) / [(Ke-g) × (1+Ke)^n]

Gordon model (single stage):
P = D_1 / (Ke - g)

Applicability checklist:

  • Has paid dividends continuously for more than 3 years
  • Stable payout ratio (>30%)
  • Strong earnings predictability
  • Usually not suitable for high-growth stocks (no dividends / low payout)

3. SOTP (Sum of the Parts)

Applicable to: diversified conglomerates.

Group value = Σ valuation of each business segment + net cash - holding-company discount

Example (a group company):
| Segment | Revenue (100m RMB) | Valuation Method | Valuation (100m RMB) |
|------|---------|---------|---------|
| Baijiu | 80 | 30x PE | 600 |
| Real estate | 50 | 0.6x PB | 120 |
| Financials | 30 | 1.0x PB | 200 |
| Total | | | 920 |
| Holding-company discount | | -15% | -138 |
| Group valuation | | | 782 |

Relative Valuation Methods

1. PE Band

Historical PE percentile analysis:
- Take the past 5 years of PE_TTM time series
- Compute the 10% / 25% / 50% / 75% / 90% percentiles
- Judge overvaluation / undervaluation from the current PE percentile

| Percentile | PE | Implied Price | Interpretation |
|------|-----|---------|------|
| 90% | 35x | 52.5 | Severely overvalued |
| 75% | 28x | 42.0 | Rich |
| 50% | 22x | 33.0 | Fair |
| 25% | 16x | 24.0 | Cheap |
| 10% | 12x | 18.0 | Severely undervalued |
| Current | 18x | 27.0 | Cheap (30th percentile) |

2. PB-ROE Matrix

Theoretical relationship: PB = (ROE - g) / (Ke - g)
Practical use: plot companies in the industry on a PB vs ROE scatter chart

| Quadrant | PB | ROE | Interpretation |
|------|-----|-----|------|
| Lower right | Low PB | High ROE | Undervalued (best buy zone) |
| Upper right | High PB | High ROE | Fair (quality premium) |
| Lower left | Low PB | Low ROE | Value trap or distressed turnaround |
| Upper left | High PB | Low ROE | Overvalued (avoid) |

3. EV/EBITDA

EV = market cap + net debt (interest-bearing debt - cash)
EBITDA = operating profit + depreciation + amortization

Advantages:
- Removes capital-structure differences (vs PE)
- Removes depreciation-policy differences
- Suitable for asset-heavy industries (telecom / energy / infrastructure)

Reference EV/EBITDA ranges by China A-share industry:
| Industry | Median | Undervalued | Overvalued |
|------|--------|------|------|
| Consumer | 15-20x | <12x | >25x |
| Technology | 12-18x | <10x | >22x |
| Energy | 6-10x | <5x | >12x |
| Utilities | 8-12x | <6x | >15x |

Valuation-Trap Detection

Top 10 Valuation Traps

#TrapDetection MethodTypical Example
1Low-PE cyclical at the peakPE is lowest when earnings are highest and about to fallCoal at 5x PE in 2021 was the top
2High-PE growth can be justifiedPEG < 1 means the growth rate supports the valuation30x PE + 40% growth = PEG 0.75
3Low-PB value destructionSustained ROE < Ke means shareholder value is being destroyedLong-term loss-making asset-heavy company
4Goodwill bombGoodwill / net assets >30% implies impairment riskUnderperforming acquisition after paying a high premium
5Accounts-receivable trapRising receivables / revenue ratio = poor revenue qualityGovernment receivables + high customer concentration
6Capitalization trapCapitalizing R&D / interest flatters profitPE doubles after true expensing
7One-off gainsLarge gap between recurring net profit and reported net profitAsset sales / government subsidies boost earnings
8Share dilutionStock options / convertible bonds reduce EPSPE should be based on diluted EPS
9Related-party transactionsBuy cheap from related parties / sell high to themProfit shifted outside the listed entity
10FX swingsHigh overseas-revenue share means large currency sensitivityRMB appreciation erodes exporter profits

Analysis Framework

Valuation-Method Selection Decision Tree

What type of company is it?
├── Mature and stable (consumer / utilities / banks)
│   ├── High dividend -> DDM
│   └── Low dividend -> DCF + PE Band
├── High growth (tech / pharma / new energy)
│   └── DCF (high-growth phase) + PEG + PS
├── Cyclical (coal / steel / nonferrous)
│   └── PB + EV/EBITDA (avoid PE)
├── Diversified conglomerate
│   └── SOTP
└── Loss-making company
    └── PS (price-to-sales) + EV/Sales

Cross-Validation

Use at least 2 valuation methods and take the middle value:
1. DCF -> intrinsic value
2. Comparable PE -> market pricing
3. If the difference >30% -> check whether assumptions are reasonable

Output Format

## Valuation Analysis: [Company Name / Code]

### Valuation Summary
| Method | Per-Share Value | Weight | Notes |
|------|---------|------|------|
| DCF | ¥32.5 | 50% | WACC=10%, g=2.5% |
| Comparable PE | ¥28.0 | 30% | Industry average 22x, EPS=1.27 |
| PB-ROE | ¥30.0 | 20% | Fair PB=2.5x |
| **Composite Target Price** | **¥30.8** | | Current price 25.0, upside +23% |

### Sensitivity Analysis
[WACC vs growth-rate matrix]

### Valuation-Trap Check
- [x] Is PE a falsely low cyclical peak? -> No
- [x] Goodwill / net asset ratio -> 12%, safe
- [x] Receivables / revenue trend -> Stable, not deteriorating
- [ ] Gap in recurring net profit -> 15% difference, subsidy dependence worth attention

### Investment Rating: Buy
Target price ¥30.8, current price ¥25.0, upside 23%

Notes

  1. DCF is highly sensitive to assumptions: a 1% change in WACC can move valuation by 20%+, so sensitivity analysis is mandatory
  2. Comparable companies must truly be comparable: same industry + same scale + same stage; do not apply leader PE multiples to small companies
  3. China A-share valuation system is unique: shell value / liquidity premium / policy premium mean US-equity standards cannot be copied directly
  4. Valuation is not a target price: markets can remain irrational for a long time, and valuation is an anchor, not a trading signal
  5. Special handling for cyclicals: use normalized earnings (mid-cycle earnings), not current earnings
  6. Not suitable for cryptocurrencies: traditional valuation frameworks do not apply to BTC / ETH; use on-chain metrics instead (see onchain-analysis)

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

ADR/H-share/A-share cross-listing premium analysis — track pricing gaps between US-listed ADRs, HK-listed H-shares, and A-shares for arbitrage signals, dual-listing valuation, and delisting risk assessment.

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

akshare

無料

AKShare financial data aggregator (18k+ stars). Free, no API key. Covers A-shares, US, HK, futures, macro, forex. Primary fallback for tushare and yfinance.

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

alpha-zoo

無料

Browse and bench the bundled alpha zoos — prebuilt cross-sectional factor libraries (Kakushadze 101, GTJA 191, Qlib 158, Fama-French / Carhart). Use when the user asks "which alphas exist", wants metadata on a named alpha, or wants to run IC/IR on a whole zoo over a universe.

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

A 股 ST/*ST 风险预测框架 — 基于最新中报/三季报或业绩预告/快报,预测下一财年是否会因营收、利润、净资产、分红不达标而被风险警示,并将新浪监管处罚记录作为独立证据面纳入风险等级。仅适用于 A 股,不预测财务造假。

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

Asset allocation theory and optimizer usage — MPT / Black-Litterman / risk budgeting / all-weather strategy, including guides for 5 optimizers and rebalancing rules.

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

Diagnose failed or underperforming backtests, locate the root cause, and fix the issue

日本語の概要は準備中です。原文の説明を表示しています。

HKUDS/Vibe-Trading3.5万2026年10月12日 更新

HKUDS のスキルをすべて見る

このスキルの問題を報告する