Back to Details

industry-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Codex adapter note

Codex adapter note

This skill is generated from 
skills/industry-research.md
so Claude Code and Codex users share one canonical workflow.
  • Treat 
    $ARGUMENTS
    as the user's request in the current Codex thread.
  • When the source mentions Claude-only surfaces such as Task, Agent, WebSearch, Bash, Read, or Write, use the closest Codex capability available in this session: subagents when available, web search when needed, shell commands for local tools, and normal file edits for workspace files.
  • Use shared project tools from 
    tools/
    in this repository. Prefer running commands from the repository root with paths like 
    python3 tools/financial_rigor.py ...
    ; if the current thread starts outside the repo, locate the actual checkout path first instead of assuming a fixed home-directory path.
  • Before starting research, run the 
    date
    command to confirm today's date; treat it as the baseline for "latest" data and state the data cutoff date in the report header. Never assume the current date from training data.
  • Preserve the research quality rules from 
    AGENTS.md
    : cross-check financial data, use exact arithmetic tools for valuation/math, and clearly label uncertainty and source gaps.
This skill is generated from 
skills/industry-research.md
so Claude Code and Codex users share one canonical workflow.
  • Treat 
    $ARGUMENTS
    as the user's request in the current Codex thread.
  • When the source mentions Claude-only surfaces such as Task, Agent, WebSearch, Bash, Read, or Write, use the closest Codex capability available in this session: subagents when available, web search when needed, shell commands for local tools, and normal file edits for workspace files.
  • Use shared project tools from 
    tools/
    in this repository. Prefer running commands from the repository root with paths like 
    python3 tools/financial_rigor.py ...
    ; if the current thread starts outside the repo, locate the actual checkout path first instead of assuming a fixed home-directory path.
  • Before starting research, run the 
    date
    command to confirm today's date; treat it as the baseline for "latest" data and state the data cutoff date in the report header. Never assume the current date from training data.
  • Preserve the research quality rules from 
    AGENTS.md
    : cross-check financial data, use exact arithmetic tools for valuation/math, and clearly label uncertainty and source gaps.

行业投资研究：产业链全景扫描 + 四大师个股分析框架

Industry Investment Research: Full Industry Chain Panoramic Scan + Four Masters' Individual Stock Analysis Framework

对 $ARGUMENTS 行业进行系统化产业链投资研究。
Conduct systematic industry chain investment research on the $ARGUMENTS industry.

研究目标

Research Objectives

从一个投资主题/逻辑链出发，完成：
  1. 验证投资逻辑链的每一个环节
  2. 绘制完整产业链全景图
  3. 扫描全球所有上市公司（A股/港股/美股/国际）
  4. 对每个细分环节的头部公司执行四大师框架分析
  5. 输出行业级投资组合配置建议
Starting from an investment theme/logic chain, complete:
  1. Verify each link of the investment logic chain
  2. Draw a complete industry chain panoramic map
  3. Scan all listed companies globally (A-shares/H-shares/U.S. stocks/international)
  4. Execute the Four Masters Framework analysis on leading companies in each segment
  5. Output industry-level investment portfolio allocation recommendations

第一步：投资逻辑链构建与验证

Step 1: Investment Logic Chain Construction and Verification

1.1 画出逻辑链

1.1 Draw the Logic Chain

用箭头链路表达从"底层趋势"到"受益标的"的因果关系，例如：
底层趋势 A
    → 导致需求 B
        → 创造瓶颈/刚需 C
            → 受益产业链 D
Express the causal relationship from "underlying trend" to "beneficiary targets" using arrow links, for example:
Underlying Trend A
    → Leads to Demand B
        → Creates Bottleneck/Rigid Demand C
            → Benefits Industry Chain D

1.2 逐环节验证

1.2 Verify Each Link

对逻辑链的每个箭头提出质疑并寻找证据：
环节核心假设验证方式数据来源
A→B搜索行业数据/预测
B→C搜索供需分析
C→D搜索实际案例/签约
Question each arrow in the logic chain and find evidence:
LinkCore AssumptionVerification MethodData Source
A→BSearch industry data/forecasts
B→CSearch supply and demand analysis
C→DSearch actual cases/contracts

1.3 寻找"已发生的验证事件"

1.3 Find "Verified Events That Have Occurred"

列出支撑该逻辑链的已签约/已落地的真实商业事件（而非预测），例如大公司的采购协议、政策文件、行业报告等。
List real commercial events that support the logic chain (not forecasts), such as procurement agreements from major companies, policy documents, industry reports, etc.

第二步：产业链全景图绘制

Step 2: Industry Chain Panoramic Map Drawing

2.1 绘制产业链结构

2.1 Draw the Industry Chain Structure

将行业拆解为上游→中游→下游→辅助环节，例如：
上游：原材料/资源开采 → 材料加工/提纯
中游：核心设备制造 → 系统集成/工程建设 → 新技术研发
下游：运营/服务 → 终端客户
辅助：检测/认证 → 维护服务 → 金融工具（ETF/信托）
Break down the industry into upstream → midstream → downstream → supporting links, for example:
Upstream: Raw Material/Resource Extraction → Material Processing/Purification
Midstream: Core Equipment Manufacturing → System Integration/Engineering Construction → New Technology R&D
Downstream: Operation/Service → End Customers
Supporting: Testing/Certification → Maintenance Services → Financial Instruments (ETF/Trust)

2.2 识别每个环节的"生意特征"

2.2 Identify "Business Characteristics" of Each Link

对每个环节标注：
环节商业模式毛利率区间竞争格局壁垒类型周期性
卖资源/卖设备/卖服务/收租垄断/寡头/充分竞争资源/牌照/技术/规模强/中/弱
Label each link with:
LinkBusiness ModelGross Profit Margin RangeCompetitive LandscapeBarrier TypeCyclicality
Selling Resources/Selling Equipment/Selling Services/Rent CollectionMonopoly/Oligopoly/Fully CompetitiveResource/License/Technology/ScaleStrong/Medium/Weak

2.3 标记"卡脖子环节"

2.3 Mark "Bottleneck Links"

识别产业链中供给最紧张、替代最难、利润率最高的环节——这些往往是最佳投资标的所在。
Identify links in the industry chain with the tightest supply, hardest substitution, and highest profit margins—these are often where the best investment targets lie.

AI研究偏见自觉：行业研究的特殊陷阱

AI Research Bias Awareness: Special Traps in Industry Research

行业研究中，AI数据偏见会以独特方式放大：
行业级偏见：
偏见类型表现应对
成熟行业偏好传统行业（银行/能源/消费）资料极多，AI分析看起来"更确定"确定性来自商业模式，不来自研报数量
新兴行业低估新行业（AI应用/合成生物等）资料少，AI分析偏保守用"终局思维"而非"当前数据"判断行业价值
龙头偏好大公司资料远多于小公司，AI天然倾向推荐龙头小公司可能有更好的风险回报比，不要因为AI分析篇幅短就忽略
上市偏好只扫描上市公司会遗漏产业链中的关键未上市玩家必须搜索未上市公司，标注"未来IPO候选"
英文偏好AI对英文资料的处理能力更强，可能低估中国/亚洲市场玩家必须同时搜索中英文信息源
产业链扫描中的反偏见措施：
  1. 对每个环节，不仅列出"AI容易找到的公司"，还要主动搜索"冷门但可能优质的标的"
  2. 对信息稀缺的小市值公司，不因分析篇幅短就降低推荐度——用核心问题（生意本质、护城河、管理层）而非报告长度来评判
  3. 在最终报告中标注每家公司的"信息充分度"（A/B/C级），让读者知道AI分析的可靠程度
AI data biases will be amplified in unique ways in industry research:
Industry-Level Biases:
Bias TypePerformanceResponse
Mature Industry PreferenceTraditional industries (banking/energy/consumption) have a huge amount of data, making AI analysis seem "more certain"Certainty comes from business models, not the number of research reports
Emerging Industry UnderestimationNew industries (AI applications/synthetic biology, etc.) have limited data, leading to conservative AI analysisUse "endgame thinking" instead of "current data" to judge industry value
Leading Company PreferenceLarge companies have far more data than small companies, so AI naturally tends to recommend leadersSmall companies may have better risk-reward ratios; don't ignore them just because AI analysis is short
Listed Company PreferenceScanning only listed companies will miss key unlisted players in the industry chainMust search for unlisted companies and mark them as "future IPO candidates"
English Language PreferenceAI has stronger processing capabilities for English materials, which may underestimate Chinese/Asian market playersMust search both Chinese and English information sources
Anti-Bias Measures in Industry Chain Scanning:
  1. For each link, not only list "companies easily found by AI" but also actively search for "obscure but potentially high-quality targets"
  2. For small-cap companies with scarce information, don't reduce recommendation just because analysis is short—judge based on core issues (essence of business, moat, management) rather than report length
  3. Mark the "information sufficiency" (Level A/B/C) of each company in the final report, so readers know the reliability of AI analysis

第三步：全球上市公司扫描

Step 3: Global Listed Company Scanning

使用 Task 工具启动后台 Agent，全面搜索该行业所有上市公司。
Use the Task tool to start a background Agent and comprehensively search all listed companies in the industry.

搜索清单

Search List

  • 美股（NYSE/NASDAQ/NYSE American）相关公司
  • A股（上海/深圳）相关公司
  • 港股相关公司
  • 其他国际市场（日本/韩国/欧洲/澳大利亚等）
  • 行业ETF
  • 关键未上市公司（可能未来IPO）
  • Relevant companies in U.S. stocks (NYSE/NASDAQ/NYSE American)
  • Relevant companies in A-shares (Shanghai/Shenzhen)
  • Relevant companies in H-shares
  • Other international markets (Japan/Korea/Europe/Australia, etc.)
  • Industry ETFs
  • Key unlisted companies (potential future IPOs)

对每家公司收集

Collect for Each Company

  • 公司名称（中英文）
  • 股票代码和交易所
  • 市值（近似）
  • 一句话描述（在产业链中的位置和作用）
  • 是否纯正标的（纯核电 vs 多元化中有核电业务）
  • 产业链所属环节
  • Company name (Chinese and English)
  • Stock code and exchange
  • Approximate market capitalization
  • One-sentence description (position and role in the industry chain)
  • Whether it is a pure target (pure nuclear power vs. nuclear power business in diversified operations)
  • Industry chain segment it belongs to

输出格式

Output Format

按产业链环节分类，每个环节一张表，包含所有扫描到的公司。 再按投资确定性分层：
  • Tier 1：大市值、纯正标的、行业龙头
  • Tier 2：中市值、纯正或高占比、细分龙头
  • Tier 3：小市值、开发阶段、高风险高弹性
  • Tier 4：多元化公司中有相关业务的大型企业
Classify by industry chain segments, with a table for each segment containing all scanned companies. Then stratify by investment certainty:
  • Tier 1: Large market capitalization, pure target, industry leader
  • Tier 2: Medium market capitalization, pure or high-proportion, segment leader
  • Tier 3: Small market capitalization, development stage, high risk and high elasticity
  • Tier 4: Large enterprises with relevant businesses in diversified operations

第四步：各环节头部公司四大师分析

Step 4: Four Masters' Analysis of Leading Companies in Each Segment

对每个产业链环节的Tier 1和Tier 2公司，执行以下分析（Tier 3/4公司简要点评即可）：
For Tier 1 and Tier 2 companies in each industry chain segment, perform the following analysis (brief comments for Tier 3/4 companies):

4.1 生意本质（段永平）

4.1 Essence of the Business (Duan Yongping)

  • 一句话定义这家公司在产业链中做什么
  • 收入结构与增速
  • 毛利率/净利率水平及趋势
  • 现金流特征
  • 追问：这是一门好生意吗？为什么？
  • One-sentence definition of what the company does in the industry chain
  • Revenue structure and growth rate
  • Gross profit margin/net profit margin level and trend
  • Cash flow characteristics
  • Follow-up Question: Is this a good business? Why?

4.2 护城河（巴菲特）

4.2 Moat (Warren Buffett)

用五类护城河评分（★1-5）：
护城河强度证据
品牌/定价权
转换成本
网络效应
规模效应
技术/牌照壁垒
追问：10年后护城河还在吗？
Score using five types of moats (★1-5):
MoatStrengthEvidence
Brand/Pricing Power
Switching Costs
Network Effects
Scale Effects
Technology/License Barriers
Follow-up Question: Will the moat still exist in 10 years?

4.3 风险（芒格）

4.3 Risks (Charlie Munger)

  • 这家公司最可能怎么失败？
  • 最坏情景下值多少钱？
  • 聪明人为什么不买？
  • How is this company most likely to fail?
  • How much is it worth in the worst-case scenario?
  • Why don't smart people buy it?

4.4 管理层（段永平+巴菲特）

4.4 Management (Duan Yongping + Warren Buffett)

  • CEO/创始人是谁？关键决策记录
  • 持股比例与利益对齐
  • 简评（A/B/C级）
  • Who is the CEO/founder? Key decision records
  • Shareholding ratio and interest alignment
  • Brief evaluation (Level A/B/C)

4.5 估值快照

4.5 Valuation Snapshot

  • 当前PE/PS/EV/EBITDA
  • 与同环节竞争对手对比
  • 简评：贵了/合理/便宜
  • Current PE/PS/EV/EBITDA
  • Comparison with competitors in the same segment
  • Brief comment: Overvalued/Fairly Valued/Undervalued

4.6 推荐度

4.6 Recommendation Rating

用★1-5标注：
  • ★★★★★ = 核心仓位候选
  • ★★★★☆ = 卫星仓位候选
  • ★★★☆☆ = 观察名单
  • ★★☆☆☆ = 高风险期权
  • ★☆☆☆☆ = 不推荐
Mark with ★1-5:
  • ★★★★★ = Core position candidate
  • ★★★★☆ = Satellite position candidate
  • ★★★☆☆ = Watchlist
  • ★★☆☆☆ = High-risk option
  • ★☆☆☆☆ = Not recommended

第五步：行业级风险评估（芒格"检查清单"）

Step 5: Industry-Level Risk Assessment (Charlie Munger's "Checklist")

5.1 系统性风险清单

5.1 Systematic Risk List

风险概率影响应对策略
投资逻辑链的某个环节被证伪
替代技术出现
政策/监管黑天鹅
需求周期性回调
估值泡沫破裂
RiskProbabilityImpactResponse Strategy
A link in the investment logic chain is falsified
Alternative technology emerges
Policy/regulatory black swan
Cyclical demand correction
Valuation bubble burst

5.2 历史类比

5.2 Historical Analogy

找到历史上类似的产业链投资主题，分析其最终结局：
  • 类比行业是什么？
  • 最终赢家是谁？（上游/中游/下游？）
  • 多数投资者赚钱了还是亏钱了？
  • 对当前行业的启示是什么？
Find similar industry chain investment themes in history and analyze their final outcomes:
  • What is the analogous industry?
  • Who were the final winners? (Upstream/midstream/downstream?)
  • Did most investors make money or lose money?
  • What are the implications for the current industry?

5.3 偏误自查

5.3 Bias Self-Check

  • 叙事偏差：故事是否太完美？
  • 锚定效应：是否被近期涨幅锚定？
  • 从众效应：是否因为"所有人都在买"？
  • Narrative bias: Is the story too perfect?
  • Anchoring effect: Are you anchored by recent price increases?
  • Herd effect: Are you buying because "everyone is buying"?

第六步：文明趋势判断（李录框架）

Step 6: Civilization Trend Judgment (Li Lu Framework)

  • 这个行业所依托的底层趋势，是"文明级范式转移"还是"阶段性热潮"？
  • 历史上最接近的技术革命类比是什么？
  • 10-20年后，这个行业的终局是什么？
  • 产业链中，哪个环节最可能出现"赢家通吃"？
  • 哪个环节最可能被颠覆？
  • Is the underlying trend that this industry relies on a "civilization-level paradigm shift" or a "phased boom"?
  • What is the closest technological revolution analogy in history?
  • What will be the endgame of this industry in 10-20 years?
  • Which link in the industry chain is most likely to have "winner-takes-all"?
  • Which link is most likely to be disrupted?

第七步：投资组合配置建议

Step 7: Investment Portfolio Allocation Recommendations

7.1 推荐组合

7.1 Recommended Portfolio

按以下结构输出：
层级仓位占比标的所属环节核心逻辑
核心仓位占主题仓位50-60%最确定、护城河最宽
卫星仓位占主题仓位25-35%弹性较大、确定性稍低
期权仓位占主题仓位5-15%高风险高回报，可以归零
ETF替代可替代以上全部不想选股的"懒人方案"
Output in the following structure:
TierPosition ProportionTargetIndustry Chain SegmentCore Logic
Core Position50-60% of theme positionMost certain, widest moat
Satellite Position25-35% of theme positionHigh elasticity, slightly lower certainty
Option Position5-15% of theme positionHigh risk and high return, may go to zero
ETF AlternativeCan replace all above"Lazy solution" for those who don't want to pick stocks

7.2 买入/卖出信号

7.2 Buy/Sell Signals

信号类型具体条件
加仓信号
减仓信号
清仓信号
Signal TypeSpecific Conditions
Increase Position Signal
Reduce Position Signal
Close Position Signal

7.3 主题仓位上限建议

7.3 Suggested Upper Limit for Theme Position

根据投资逻辑链的确定性和风险程度，建议该主题占总仓位的上限百分比。
Based on the certainty and risk level of the investment logic chain, suggest the upper limit percentage of this theme in the total position.

第八步：综合决策备忘录

Step 8: Comprehensive Decision Memo

行业总评表

Industry Overall Evaluation Table

维度结论信心度
投资逻辑链（验证程度）
最佳环节（段永平"对的生意"）
最宽护城河（巴菲特）
最大风险（芒格）
文明趋势定位（李录）
整体估值水平
DimensionConclusionConfidence Level
Investment Logic Chain (Verification Level)
Best Segment (Duan Yongping's "Right Business")
Widest Moat (Warren Buffett)
Biggest Risk (Charlie Munger)
Civilization Trend Positioning (Li Lu)
Overall Valuation Level

四位大师模拟点评

Simulated Comments from the Four Masters

用引用格式，模拟四位大师对该行业投资机会的点评。
Use quote format to simulate comments from the four masters on the industry investment opportunity.

输出要求

Output Requirements

  1. 所有分析必须有数据支撑，附数据来源
  2. 使用 Markdown 表格呈现关键数据
  3. 产业链全景图用代码块的文本图表示
  4. 每个环节至少分析2-3家头部公司
  5. 全球公司扫描要尽可能完整（A股/港股/美股/国际）
  6. 最终将完整报告写入 
    ~/[行业名]产业链投资研究报告.md
  7. 结论要明确，给出具体的标的、仓位和价格区间建议
  8. 每个分析模块末尾有对应大师的"追问"
  1. All analysis must be supported by data, with data sources attached
  2. Use Markdown tables to present key data
  3. Represent the industry chain panoramic map with a text diagram in a code block
  4. Analyze at least 2-3 leading companies for each segment
  5. Make the global company scan as complete as possible (A-shares/H-shares/U.S. stocks/international)
  6. Finally write the complete report to 
    ~/[Industry Name] Industry Chain Investment Research Report.md
  7. Conclusions must be clear, with specific recommendations for targets, positions, and price ranges
  8. Include a "follow-up question" from the corresponding master at the end of each analysis module

数据抽检（准出流程）

Data Sampling Inspection (Approval Process)

报告写入后，执行数据抽检，通过方可发布：
bash
undefined
After writing the report, perform data sampling inspection; it can be published only if passed:
bash
undefined

Step 1 — 提取抽检清单（15%随机抽样）

Step 1 — Extract inspection list (15% random sampling)

python3 tools/report_audit.py extract
--report <报告文件路径>
python3 tools/report_audit.py extract
--report <Report File Path>

Step 2 — 对清单每项从可靠信源取数（参见 skills/financial-data.md）

Step 2 — Retrieve data for each item in the list from reliable sources (see skills/financial-data.md)

Step 3 — 输出准出/打回判决

Step 3 — Output approval/rejection verdict

python3 tools/report_audit.py verdict
--results '<填好的JSON>'
--report <报告文件名>

**【准出】** 全部通过 → 报告可发布；**【打回】** 有不通过 → 修正后重审。
python3 tools/report_audit.py verdict
--results '<Filled JSON>'
--report <Report File Name>

**【Approval】** All passed → Report can be published; **【Rejection】** Any failed → Revise and re-review.