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.

Quality

58%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./agent/src/skills/valuation-model/SKILL.md

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

Year	Revenue (100m RMB)	EBIT (100m RMB)	FCFF (100m RMB)	Growth
2026E	120	24	18	+15%
2027E	138	28	21	+15%
2028E	155	31	24	+12%
2029E	170	34	26	+10%
2030E	182	36	28	+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:

Industry	WACC Range	Reference β
Consumer	8-10%	0.8-1.0
Technology	10-13%	1.2-1.5
Financials	7-9%	1.0-1.2
Cyclicals	9-12%	1.0-1.4
Utilities	6-8%	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

#	Trap	Detection Method	Typical Example
1	Low-PE cyclical at the peak	PE is lowest when earnings are highest and about to fall	Coal at 5x PE in 2021 was the top
2	High-PE growth can be justified	`PEG < 1` means the growth rate supports the valuation	30x PE + 40% growth = PEG 0.75
3	Low-PB value destruction	Sustained `ROE < Ke` means shareholder value is being destroyed	Long-term loss-making asset-heavy company
4	Goodwill bomb	Goodwill / net assets >30% implies impairment risk	Underperforming acquisition after paying a high premium
5	Accounts-receivable trap	Rising receivables / revenue ratio = poor revenue quality	Government receivables + high customer concentration
6	Capitalization trap	Capitalizing R&D / interest flatters profit	PE doubles after true expensing
7	One-off gains	Large gap between recurring net profit and reported net profit	Asset sales / government subsidies boost earnings
8	Share dilution	Stock options / convertible bonds reduce EPS	PE should be based on diluted EPS
9	Related-party transactions	Buy cheap from related parties / sell high to them	Profit shifted outside the listed entity
10	FX swings	High overseas-revenue share means large currency sensitivity	RMB 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

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

Repository: HKUDS/Vibe-Trading
Path: agent/src/skills/valuation-model/SKILL.md
Commit: 8643fcd

Last updated: 2 days ago
First committed: 4 months ago

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.