藏在《投资最重要的事》里的Skill

Skill.md

投资最重要的事

霍华德·马克斯在《投资最重要的事》（原书名《The Most Important Thing》）中提出了一套可执行的判断框架。这个 Book2Skills 页面把它转化为可直接调用的 AI Skill：先识别问题类型，再按维度检查证据、风险信号和输出格式，帮助用户把书中的方法用于真实决策。

核心框架

• 适用场景：投资, 风险管理, 价值投资, 市场周期
• 输入方式：给出你要判断的对象、目标、约束和已有数据。
• 输出方式：Skill 会按书中方法生成诊断、检查清单、风险信号和下一步建议。

支持的问题类型

• 诊断一个真实案例是否符合书中原则。
• 比较多个方案并给出取舍理由。
• 找出风险、反模式或缺失信息。
• 把抽象原则转成可执行步骤。

完整 Skill 指令

以下保留完整 Skill 指令，供 agent 调用时使用。

# The Most Important Thing — Risk Investing Skill

**Knowledge source:** *The Most Important Thing* by Howard Marks.

## Overview

Use this skill to evaluate investment opportunities, portfolio posture, market temperature, risk compensation, and contrarian decisions through Howard Marks's multi-factor framework. It supports investors asking whether to buy, hold, sell, wait, or become more defensive.

## When to Use This Skill

Use this skill when the user asks:
- "Is this asset worth buying?"
- "What risks am I missing?"
- "Where are we in the cycle?"
- "Should I be contrarian here?"
- "Is the price attractive relative to value?"
- "How defensive should I be?"

## Core Principle

There is no single most important thing. Superior investing requires second-level thinking across value, price, risk, cycles, psychology, contrarianism, patience, humility, and defense, with survival and risk control placed ahead of aggressive return seeking.

## Workflow Inventory

| Workflow | User question pattern | Inputs | Steps | Output | Independent trigger? | Distinct references? | Triage score | Should be subskill? | Reason |
|---|---|---|---|---|---|---:|---:|---|---|
| Investment attractiveness | "Should I buy X?" | Asset, price, value estimate, risk, market context | Apply value, risk, cycles, psychology, margin | Buy/watch/avoid verdict | Yes | Yes | 4 | No | All dimensions are coupled in Marks's "no single thing" principle. |
| Risk diagnosis | "What risks?" | Position, leverage, price, assumptions | Identify permanent-loss risks and unpriced risks | Risk register | Yes | Yes | 3 | No | Same final investment report. |
| Cycle/temperature assessment | "Where are we?" | Market indicators, sentiment, valuations | Assess pendulum, psychology, risk appetite | Posture recommendation | Yes | Yes | 3 | No | Feeds buy/sell aggressiveness. |
| Contrarian opportunity | "Should I go against crowd?" | Consensus, price, fundamentals, forced selling | Test bargain source and humility | Contrarian thesis check | Yes | Yes | 3 | No | Must be combined with value and risk. |

## Architecture Justification

Several workflows score as subskill candidates, but the book explicitly argues that investing judgment fails when any "most important thing" is isolated. A single-file architecture is justified because every supported query must combine price-value, risk, psychology, cycle position, and humility.

## DIMENSION 1: Second-Level Thinking

**The Rule:** Outperformance requires thinking differently and better than the consensus.

### Key questions to ask:
- What does the consensus believe?
- What would a first-level answer say?
- What must be true for the consensus to be wrong?
- Is the user's view different for a reason or just different?

### Decision criteria / Checklist:
- Identify consensus expectations.
- State variant perception.
- Test whether evidence supports that variant.
- Avoid contrarianism without value support.

### Warning signals:
- "Good company, therefore good investment."
- Being different only for identity or excitement.
- Ignoring what is already in the price.

### Agent instruction:
Before giving any investment verdict, write the first-level view and the second-level counterview.

## DIMENSION 2: Price, Value, and Bargains

**The Rule:** Price determines return; value only matters if the price paid is attractive.

### Key questions to ask:
- What is conservative intrinsic value?
- What expectations are embedded in the price?
- Why does this bargain exist?
- Is the discount large enough for uncertainty?

### Decision criteria / Checklist:
- Estimate value conservatively.
- Compare price to value.
- Identify bargain source: neglect, forced selling, misunderstanding, fear.
- Require margin for error.

### Warning signals:
- Paying any price for quality.
- Calling something cheap because it has fallen.
- No explanation for why the market is mispricing it.

### Agent instruction:
For any buy question, refuse to evaluate attractiveness without price versus value.

## DIMENSION 3: Risk Recognition and Control

**The Rule:** Risk is the probability of permanent loss, not mere volatility, and it often rises when people believe risk is absent.

### Key questions to ask:
- What can cause permanent impairment?
- What risk is hidden by recent good performance?
- Is risk compensated by price?
- What happens under adverse scenarios?

### Decision criteria / Checklist:
- Separate volatility from impairment.
- Identify leverage, overpayment, concentration, fragility, and ignorance.
- Demand compensation for risk.
- Favor survival in uncertain environments.

### Warning signals:
- "There is no risk."
- Rising prices treated as proof of safety.
- Leverage or optimism covering thin margins.

### Agent instruction:
Always produce a risk register before a return thesis.

## DIMENSION 4: Cycles, Psychology, and Contrarianism

**The Rule:** Markets swing between greed and fear; good posture depends on where we stand.

### Key questions to ask:
- Are investors euphoric, fearful, complacent, or capitulating?
- Are valuations, credit, and narratives stretched?
- Is the user being influenced by envy, fear, or conformity?
- Is a contrarian action supported by value and risk?

### Decision criteria / Checklist:
- Assess market temperature.
- Identify pendulum position.
- Counter emotional extremes.
- Act aggressively only when odds and price are favorable.

### Warning signals:
- "This time is different."
- Fear of missing out.
- Capitulation after price declines.
- Crowded certainty.

### Agent instruction:
For market-timing or posture questions, recommend calibration rather than prediction.

## DIMENSION 5: Patience, Humility, and Defense

**The Rule:** Most of the time, the best action is patient defense until odds are favorable.

### Key questions to ask:
- What is knowable and what is not?
- Is there pressure to act without a pitch?
- Does the portfolio survive bad outcomes?
- Are expectations reasonable?

### Decision criteria / Checklist:
- Avoid forecasts that require precision.
- Prefer patient opportunism.
- Emphasize defensive positioning when risk compensation is poor.
- Separate luck from skill in past outcomes.

### Warning signals:
- Always needing to be fully invested.
- Mistaking a lucky outcome for skill.
- Ignoring unimaginable scenarios.

### Agent instruction:
When evidence is insufficient, recommend watchlist, data needs, or defensive posture instead of forcing a verdict.

## Query Response Framework

### Query Type 1: Should I buy or add?
1. State first-level and second-level views.
2. Analyze value vs price.
3. Build risk register.
4. Assess cycle/psychology.
5. Give buy/watch/avoid verdict with missing data.

### Query Type 2: What risks am I missing?
1. Identify permanent-loss scenarios.
2. Separate volatility from impairment.
3. Test whether price compensates for risk.
4. Recommend controls or position limits.

### Query Type 3: Market posture
1. Assess pendulum and market temperature.
2. Identify emotional influences.
3. Recommend offensive/defensive calibration.

## Output Format

```markdown
## Howard Marks Investment Judgment
**Asset / Decision:** ...
**Verdict:** Buy / Watch / Avoid / Hold / Reduce / Needs data

### First-Level vs Second-Level View
...

| Dimension | Finding | Implication |
|---|---|---|

### Risk Register
...

### Action Discipline
...

### Citations
...

Critical Reminders

1. No asset is so good it cannot become overpriced.
2. Risk often rises when people believe risk is gone.
3. Contrarianism requires evidence, not reflex.
4. Forecasting humility is part of the method.
5. Defense and survival come before upside maximization.

CITATION RULES

Every substantive Marks-method claim must include a citation to the original text.

Quote files:

• market-cycles-quotes.md — cycles, pendulum, negative influences, market temperature, luck, expectations, survival, and bad lessons from good times.
• risk-quotes.md — risk definition, invisibility, high prices, risk control, defensive investing, and down-market outperformance.
• second-level-thinking-quotes.md — second-level thinking, efficiency limits, unconventional behavior, and thinking differently.
• value-investing-quotes.md — price/value, bargains, patience, contrarianism, humility, and the no-single-thing principle.

Citation format:

"Author's exact words here."

— The Most Important Thing, cited excerpt

Anchor mapping:

• market-cycles-quotes.md: #everything-cycles, #pendulum-swings, #six-negative-influences, #where-we-stand, #luck-not-skill, #reasonable-expectations, #survival-first, #good-times-teach-bad-lessons
• risk-quotes.md: #risk-not-volatility, #risk-invisible, #high-prices-create-risk, #no-risk-is-the-risk, #rising-prices-create-risk, #risk-control-not-avoidance, #biggest-losses-unimaginable, #defensive-investing, #outperformance-down-markets
• second-level-thinking-quotes.md: #second-level-defined, #first-vs-second, #outperformance-requires-difference, #be-wrong-look-foolish, #market-efficiency-limitations, #inefficiency-sources, #unconventional-behavior, #thinking-differently
• value-investing-quotes.md: #second-level-thinking, #no-asset-so-good, #price-determines-return, #value-investing-defined, #finding-bargains, #patient-opportunism, #contrarianism-essential, #this-time-is-never-different, #know-what-you-dont-know, #no-single-most-important-thing

Rules:

• Cite one anchor for each major section.
• Never invent quote text.
• For modern securities, label conclusions as applications of Marks's framework.