Margin of Safety - Prompt de Análisis IA

Use this Warren Buffett rule prompt to apply “Insistir en el Margen de Seguridad” to a specific company. It turns a vague opinion into a repeatable checklist: what facts you must verify, which assumptions matter most, what would invalidate the thesis, and the common misreads that create false certainty. Expect a written output you can save: a thesis summary, key risks, and next-step questions for filings and earnings calls. If a claim matters, require primary-source citations before you act. Educational only — not investment advice.

Prompt completo

Eres un analista de valoración especializado en el principio de "Margen de Seguridad" de Warren Buffett. Tu tarea es calcular el valor intrínseco de {Nombre de la Empresa} y determinar si el precio actual ofrece un margen de seguridad adecuado.

## Marco de Análisis
### 1. Estimación de Valor Intrínseco — Método 1: Análisis DCF
- Proyecta flujos de caja libres para los próximos 10 años
- Usa supuestos de crecimiento conservadores (por debajo del consenso de analistas)
- Tasa de crecimiento terminal: no mayor que el crecimiento del PIB a largo plazo
- Tasa de descuento apropiada al riesgo

### 2. Método 2: Ganancias del Propietario
- Calcula las ganancias del propietario según la definición de Buffett
- Aplica un múltiplo razonable

### 3. Método 3: Valoración Basada en Activos
- Valor de liquidación y valor de reemplazo
- Activos intangibles y su valor real

### 4. Análisis de Margen de Seguridad
- Precio actual vs. valor intrínseco (promedio de los 3 métodos)
- ¿El margen es suficiente? (Buffett típicamente busca >25%)
- Pruebas de estrés: ¿Qué pasa si las estimaciones están equivocadas en 20%?

Proporciona valoraciones específicas con números concretos.

Related reading (close the loop)

Pick one path below to turn the output into a checkable, repeatable decision policy.

Educational only. Verify facts with primary sources and apply your own constraints.

ℹ️Este contenido solo está disponible en chino e inglés por el momento.

Basic Questions

Why use three valuation methods? Isn't one enough?
Valuation is art, not science — each method has blind spots:

📊 DCF (Discounted Cash Flow):
- Strength: Theoretically most correct
- Blind spot: Extremely sensitive to growth rate assumptions

📊 Owner Earnings Method:
- Strength: Buffett's preferred method, focuses on real profits
- Blind spot: Not suitable for capital-intensive industries

📊 Asset-Based Method:
- Strength: Provides "worst case" floor
- Blind spot: Ignores earning power

Cross-validating three methods to get a "value range" is more reliable than a single number.

Usage Tips

How to make AI valuation more accurate?
✅ Key: Provide real financial data

AI valuation quality entirely depends on input data quality. Recommendations:

1. Extract 5-10 years of key data from annual reports and append to the prompt:
- Revenue, net income, free cash flow
- ROE, ROIC
- CapEx, depreciation
- Share count (check for dilution)

2. Explicitly tell AI your reasonable assumption ranges:
- "Assume revenue growth of 8-12% for the next 5 years"
- "Use 10% discount rate"

3. Ask AI to show sensitivity analysis table, not a single number

Getting started

Does this prompt give investment advice or buy/sell calls?
No. It is a research helper that turns your thinking into checkable inputs and constraints: what evidence you must verify, what would prove the thesis wrong, and what common misreads to avoid. Treat the output as a draft, not a signal. Validate every material number against primary sources (filings, earnings releases, investor presentations, transcripts), and do not act unless you can write down (1) position-size limits and (2) explicit invalidation triggers.
What inputs should I provide for a reliable result?
At minimum: a 1-sentence business model summary, your current thesis (why it wins/loses), time horizon, and risk constraints; a valuation/price range; and the latest financial statements (profit quality, cash flow, debt/liquidity). Add context that reduces hallucinations: the exact filing period, known one-offs, key competitors, and what you do NOT know yet. If an input is missing, label it as missing evidence instead of letting the model guess.

Validation and boundaries

How do I validate the output?
Validate falsifiable claims one by one. Rewrite each key statement into something you can check: the metric, the period, and the source. Numbers must match filings; management claims must be traceable to transcripts/guidance; and “moat” claims need observable evidence (pricing power, retention, switching costs, cost structure). Anything you cannot verify becomes a follow-up task, not a decision trigger. If the model cites dates, confirm they are not beyond its knowledge cutoff.
When should I NOT act on the output?
If you cannot write down invalidation triggers, a position-size cap, or primary-source evidence for the key claims behind “Insistir en el Margen de Seguridad”, do not act. The safer move is usually to reduce size, slow down, and schedule the next review.

Más prompts de reglas

Explora otros principios de inversión de este maestro.