xbtlin/ai-berkshire Source: https://github.com/xbtlin/ai-berkshire English | 中文 # AI Berkshire - Value Investing Research Framework for the AI Era > “Price is what you pay, value is what you get.” — Warren Buffett > > Redefining the depth and efficiency of investment research with AI. AI Berkshire is a collection of investment research Skills built on Claude Code (https://claude.ai/code). It systematizes and structures the methodologies of four value investing masters — Warren Buffett, Charlie Munger, Duan Yongping, and Li Lu — enabling professional-grade investment research through AI agents. One person + Claude = a full investment research team. — ## Real Track Record > Not just theory. This framework is backed by real money. ### 2024 Full Year Return: +69.29% ### 2025 YTD Return: +66.38% ### Comparison with Major Indices | Metric | 2024 Full Year | 2025 YTD | |——|–––––|–––––| | This Framework Real Account | +69.29% | +66.38% | | Hang Seng Index | +17.67% | +27.77% | | S&P 500 | +23.31% | +16.39% | | CSI 300 | +14.68% | +17.66% | | Nasdaq | +28.64% | +20.36% | 2024 Excess Return: Beat S&P 500 by 46 percentage points, beat Hang Seng by 52 percentage points 2025 Excess Return: Beat S&P 500 by 50 percentage points, beat Hang Seng by 39 percentage points Cumulative real account return over two years exceeded 1.46 million RMB, significantly outperforming major global indices for two consecutive years. > Disclaimer: Past performance does not guarantee future results. Screenshots are from real Futu Securities accounts. — ## Why Not Just Ask AI Directly? You could certainly ask Claude: “Help me analyze whether Pinduoduo is worth buying.” You’d get a balanced analysis full of “on one hand… on the other hand…” ending with “Investing involves risks, please make your own judgment.” That kind of analysis looks reasonable but is useless for making decisions. AI Berkshire doesn’t solve the “can it analyze” problem — it solves the analysis quality and decision discipline problem. Here are the core differences: ### 1. Forces a Conclusion, No Fence-Sitting Ask AI directly, you get a wishy-washy analysis that tries to please everyone. AI Berkshire forces output: Pass / Fail / Grey Zone, with specific price ranges and tiered recommendations. > Typical AI response: “Pinduoduo has growth potential but also faces competitive pressure. Investors need to weigh…” > > AI Berkshire output: > | Strategy | Recommendation | Price Range | > |——|——|———| > | Aggressive | Can build 20% position at current price | $95-105 | > | Conservative | Wait for buyback policy clarity before building | 85-95 | > | Defensive | Fails 10-year certainty test, stay on sidelines | — | > > **Mirror Test**: Can't articulate the thesis in 5 sentences = don't buy. No exceptions. ### 2. Multi-Master Perspectives in Tension, Not Single Analysis It's not as simple as "analyze using Buffett's methods." Four perspectives produce **real contradictions and tensions** — Using Pinduoduo as an example: - **Duan Yongping** (Business Model): Good business, C2M model hard to replicate → Score 3.7/5 - **Buffett** (Financial Valuation): PE excl. cash only 6.3x, cash cow → Score 4.4/5 - **Munger** (Inversion Thinking): Moat is shallower than you think, Douyin reached 4 trillion GMV in 3 years → Score 3.5/5 - **Li Lu** (Long-term Certainty): Management culture has hidden risks, uncertain beyond 10 years → Score 2.0/5 **Buffett says "truly cheap", Li Lu says "uncertain = don't buy"** — this kind of conflict is the real state of investment decisions. A single prompt cannot create this multi-perspective tension, which is exactly the key to avoiding blind spots. ### 3. Structured Anti-Bias Mechanisms The most dangerous thing about AI isn't giving a wrong answer, but giving an **answer that looks right but can't withstand scrutiny**. AI Berkshire embeds multiple layers of "fraud prevention" in its process: | Mechanism | Problem Addressed | Example | |------|------------|------| | **Information Richness Rating (A/B/C)** | Prevents illusion that "more data = higher certainty" | Pop Mart rated B: limited data, flagged confidence levels for estimated metrics | | **Munger-style Inversion Check** | Forces thinking about failure scenarios | "What would kill Pinduoduo?" → List 5 scenarios with probabilities | | **Fast Veto List** | 8 red lines for instant rejection | Management integrity issue → Reject immediately, no matter how cheap | | **Contrarian Check** | Avoids thinking the same as the market | "Why are smart people shorting?" → Uncovers overlooked risks | | **"I Don't Know" Principle** | Better to say "I don't know" | When data is insufficient, mark as "Grey Zone" instead of pretending certainty | ### 4. Financial Data Precision LLM mental math is unreliable. A single decimal error in PE or mixing up CNY and HKD for market cap can lead to wrong investment decisions. **Real case**: When analyzing Tencent, market cap data from different sources had units of "HKD billions" and "CNY billions". AI Berkshire's handling: `bash # Market cap manual verification: price × total shares, compare with reported data python3 tools/financial_rigor.py verify-market-cap \ --price 510 --shares 9.11e9 --reported 4.65e12 --currency HKD # ✅ Verification passed, deviation only 0.08% ` All calculations use Python `decimal.Decimal` (precise decimal), never `float`. Key data cross-verified from at least 2 independent sources. ### 5. Reproducible Research Process Asking AI directly gives different format, depth, and coverage every time — today's Tencent analysis has a moat score, but tomorrow's Meituan analysis might forget it. AI Berkshire ensures: **Same input → Consistent structure, consistent depth output**. This means you can: - Compare 7 companies horizontally with identical scoring standards - Re-analyze the same company 6 months later and compare changes directly - Align research results across team members > Real output — 7 companies screened with the same unified Checklist: > > | Company | Pass? | Circle of Competence | Great Business | Moat | Management | Margin of Safety | Overall | > |------|:-----:|:------:|:------:|:------:|:------:|:-------:|:----:| > | Moutai | ✅ Pass | ★★★★★ | ★★★★★ | ★★★★★ | ★★★☆☆ | ★★★★☆ | 4.7 | > | Tencent | ✅ Pass | ★★★★☆ | ★★★★★ | ★★★★★ | ★★★★★ | ★★★★☆ | 4.7 | > | NVIDIA | ✅ Conditional | ★★★★☆ | ★★★★★ | ★★★★★ | ★★★★★ | ★★★☆☆ | 4.3 | > | Meituan | ✅ Conditional | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | 4.0 | > | Kuaishou | ✅ Conditional | ★★★☆☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★★ | 4.0 | > | PDD | ❓ Grey | ★★★★☆ | ★★★★☆ | ★★★☆☆ | ★★★☆☆ | ★★★★★ | 3.8 | > | Pop Mart | ❓ Grey | ★★★☆☆ | ★★★★☆ | ★★★★☆ | ★★★★★ | ★★★☆☆ | 3.7 | ### 6. Multi-Agent Parallel = Multiplied Research Depth `/investment-team` launches 4 independent agents **simultaneously** to research one company. Each agent searches the web, cross-validates data, and draws independent conclusions. This isn't splitting one prompt into four parts — it's 4 "analysts" each doing a complete research, then the Team Lead synthesizes. A person asking AI directly has one context window. 4 agents in parallel means 4x search volume, 4x information sources, 4 independent perspectives. `` ┌─────────────────────────────────────────────┐ │ Team Lead (You) │ │ Coordination · Synthesis & Judgment │ ├──────┬──────┬──────────┬───────────┤ │ Agent 1 │ Agent 2 │ Agent 3 │ Agent 4 │ │ Business Model│Financial Val.│Industry Comp.│ Risk Mgmt │ │ Duan Yongping │ Buffett View │ Munger View │ Li Lu View │ └──────┴──────┴──────────┴───────────┘ ↓ Parallel research, real-time progress updates ↓ Final comprehensive report `` ### In a Nutshell > **Asking AI directly gets you "analysis that looks right." Using AI Berkshire gets you "an investment research report you can make decisions on."** --- ## Overall Architecture > Source: [`assets/architecture.mmd`](assets/architecture.mmd) (Mermaid editable source) **Three-layer design philosophy**: - **Skill Layer**: Abstract "what you want to do" into 16 clear entry points — deep research, earnings analysis, industry screening, portfolio management, thinking tools — choose by scenario - **Agent Layer**: Inside each skill, 4 agents work in parallel — each independently searches, judges, and challenges; Team Lead synthesizes - **Tool Layer**: Precise calculation, real-time retrieval, report inspection — ensures every report's data rigor is verifiable --- ## Skills Overview (16 total) ### 🔬 Deep Research | Skill | Use | Best For | |-------|------|---------| | [`/investment-research`](skills/investment-research.md) | Four-master comprehensive deep analysis | Full-spectrum investment research on a listed company | | [`/investment-team`](skills/investment-team.md) | Multi-agent parallel research team | 4 agents in parallel — fastest, most comprehensive | | [`/management-deep-dive`](skills/management-deep-dive.md) | Deep dive into management | "Buying stocks is buying people" — when management is the core variable | | [`/private-company-research`](skills/private-company-research.md) | Deep research on unlisted companies | Research information-scarce unlisted companies like Ant Group, SpaceX | | [`/deep-company-series`](skills/deep-company-series.md) | 8-article series dissecting one company | In-depth series of ~120,000 characters, from cognitive reset to decision closure | ### 📊 Earnings Analysis | Skill | Use | Best For | |-------|------|---------| | [`/earnings-review`](skills/earnings-review.md) | In-depth earnings reading (primary sources) | Read only original filings, no secondary reports — read annual reports like Buffett | | [`/earnings-team`](skills/earnings-team.md) | Earnings reading team + WeChat article publishing | Four masters parallel interpretation → editing/polishing → reader review → publishable article | ### 🏭 Industry Screening | Skill | Use | Best For | |-------|------|---------| | [`/industry-research`](skills/industry-research.md) | Full industry chain panorama | Study all investment opportunities in an industry (sliced by chain segments) | | [`/industry-funnel`](skills/industry-funnel.md) | Industry funnel screening | Full market → rough screen ≤10 → final 3 deep dives | | [`/quality-screen`](skills/quality-screen.md) | Quality screening (7 hard criteria) | Quickly eliminate non-first-class companies; supports batch screening by stock/sector/index/theme | | [`/investment-checklist`](skills/investment-checklist.md) | Buffett pre-purchase checklist | Six-gate quick screen; decide in 10 minutes if worth deeper research | ### 📈 Portfolio Management | Skill | Use | Best For | |-------|------|---------| | [`/portfolio-review`](skills/portfolio-review.md) | Portfolio management & optimization | Upgrade from "researching companies" to "managing a portfolio" — position sizing, concentration, rebalancing | | [`/thesis-tracker`](skills/thesis-tracker.md) | Investment thesis tracking | Discipline system after purchase: continuously monitor whether the thesis is being falsified | | [`/news-pulse`](skills/news-pulse.md) | Quick attribution of stock price anomalies | Understand "what happened" in 10 minutes when stock surges/plunges | ### 🧠 Thinking Tools | Skill | Use | Best For | |-------|------|---------| | [`/dyp-ask`](skills/dyp-ask.md) | Duan Yongping Q&A | Think about any problem the Duan Yongping way — business, investing, life | | [`/financial-data`](skills/financial-data.md) | Financial data acquisition & cross-validation standards | Ensure key data from 2 independent sources; alert if error >1% | --- ## Quick Start ### 1. Install Claude Code `bash npm install -g @anthropic-ai/claude-code ` ### 2. Install Skills Copy the `.md` files from the `skills/` directory to your Claude Code commands directory: `bash # Clone repository git clone https://github.com/xbtlin/ai-berkshire.git # Copy skills to Claude Code global commands directory cp ai-berkshire/skills/*.md ~/.claude/commands/ ` ### 3. Usage Call directly in Claude Code: `bash # Deep Research /investment-research Tencent /investment-team Meituan /management-deep-dive Wang Xing Meituan /private-company-research SpaceX /deep-company-series Pinduoduo # Earnings Analysis /earnings-review Tencent 2025Q4 /earnings-team PDD 2025 Annual Report # Industry Screening /industry-research Nuclear Power /industry-funnel AI Computing /quality-screen Hang Seng Index Constituents /investment-checklist Moutai, NVIDIA, Apple # Portfolio Management /portfolio-review Tencent30%, Meituan20%, Moutai20%, Cash30% /thesis-tracker Pinduoduo /news-pulse Tencent # Thinking Tools /dyp-ask Where exactly is Pinduoduo's moat? ` --- ## Detailed Skill Descriptions ### 1. `/investment-research` — Four-Master Comprehensive Analysis The most comprehensive single-company deep research framework. Executed sequentially through seven modules: `Data Collection → Business Essence (Duan Yongping) → Moat (Buffett) → Inversion Thinking (Munger) → Management Assessment (Duan Yongping + Buffett) → Civilization Trend (Li Lu) → Valuation & Margin of Safety ` **Core Features**: - AI research bias awareness mechanism (A/B/C information richness rating) - Multi-source cross-validation of key data (manual market cap verification, at least 2 independent sources) - "Follow-up questions" from four masters throughout the text - Three-scenario valuation (Optimistic/Bearish/Baseline) + Reverse DCF **Output Example Excerpt**: > #### Comprehensive Decision Memo > > | Dimension | Conclusion | Confidence | > |------|------|--------| > | Business Quality (Duan Yongping) | Excellent: Platform business, bilateral network effects, near-zero marginal cost | ★★★★★ | > | Moat (Buffett) | Wide and widening: network effects + switching costs + economies of scale triple stack | ★★★★☆ | > | Management (Duan Yongping + Buffett) | Excellent: Founder-led, strong capital allocation discipline | ★★★★☆ | > | Biggest Risk (Munger) | Regulatory policy uncertainty; new business losses dragging overall profit | ★★★☆☆ | > | Civilization Trend (Li Lu) | Aligns with digital consumption trend, but not a "civilization-level paradigm shift" | ★★★★☆ | > | Valuation (Buffett + Duan Yongping) | Current PE 18x, below historical median — some margin of safety | ★★★★☆ | > > **Duan Yongping**: "The essence of this business is connecting consumers and merchants, making money from efficiency improvements. The hallmark of a good business: more users → more merchants; more merchants → more users. Once the flywheel starts, it's hard to stop." > > **Munger**: "Think in reverse — if this company disappeared tomorrow, what would users and merchants do? If the answer is 'they'd quickly find a substitute,' the moat isn't deep enough. If the answer is 'life would become very inconvenient,' that's worth paying attention to." --- ### 2. `/investment-team` — Multi-Agent Research Team Launches 4 AI agents in parallel, simulating a real research team collaboration. Each agent independently searches, analyzes, and scores; the Team Lead then synthesizes. **Output Example Excerpt**: > #### One-Sentence Conclusion > Meituan is the absolute leader in China's local services, with multi-layered network effects as its moat. Current valuation is near historical lows, offering significant long-term investment value. Recommend building positions on dips. > > #### Four-Dimensional Scoring Table > > | Dimension | Framework | Score | Core Judgment | > |------|------|------|----------| > | Business Model & Moat | Duan Yongping | ★★★★☆ | Strong bilateral network effects; food delivery + in-store form a flywheel | > | Financial & Valuation | Buffett | ★★★★☆ | Core business margins improving; valuation at historical lows | > | Industry & Competition | Munger | ★★★☆☆ | Douyin encroaching on in-store business; competitive landscape could worsen | > | Risk & Management | Li Lu | ★★★★☆ | Wang Xing has excellent strategic vision, but new business cash burn is a concern | > > **Overall Score: 3.8 / 5** > > #### Investment Recommendation > > | Strategy | Recommendation | Price Range (HKD) | > |------|------|---------------| > | Aggressive | Can build 30% position at current price | 120-140 | > | Conservative | Wait for pullback to 100-110 before building | 100-120 | > | Defensive | Wait for quarterly results to confirm margin trajectory | <100 | --- ### 3. `/investment-checklist` — Buffett Pre-Purchase Checklist Six-gate quick screen to decide in 10 minutes whether a company deserves deeper research: ` Gate 1: Circle of Competence (Do I understand it?) ↓ Pass Gate 2: Great Business (What are the economic characteristics?) ↓ Pass Gate 3: Moat (How deep are the competitive advantages?) ↓ Pass Gate 4: Management (Can I trust them?) ↓ Pass Gate 5: Margin of Safety (Is the price cheap?) ↓ Pass Gate 6: Decision Discipline (Is it reason or FOMO?) ↓ Pass ✅ Mirror Test ` **Supports multi-company comparison** — screen multiple names at once: `/investment-checklist Tencent, Alibaba, Meituan, Pinduoduo ` **Output Example Excerpt**: > #### Mirror Test > > "I'm buying Tencent at 380 HKD because: > 1. The essence of this business is a **social network + digital content platform**, I understand it; > 2. Its moat is **the social graph of 1.2 billion users**, and it's widening; > 3. Management **Pony Ma is low-key, pragmatic, and excellent at capital allocation**, trustworthy; > 4. The current price represents a **20% discount** to intrinsic value, offering some margin of safety; > 5. Even if I'm wrong, downside is manageable because **over 200B HKD in net cash on the balance sheet and strong gaming cash flow**." > > ✅ Passes Mirror Test > > **Can't articulate in 5 sentences = don't buy. No exceptions.** --- ### 4. `/industry-research` — Full Industry Chain Panorama Starting from an investment theme, completes a full industry chain study: ` Build investment logic chain → Full industry chain map → Global listed company scan → Four-master analysis of leading players in each segment → Portfolio allocation recommendation ` **Output Example Excerpt**: > #### Investment Logic Chain: Nuclear Power > > Underlying Trends: AI data center electricity demand surge + carbon neutrality goals > → Leads to: Explosive demand for stable, clean baseload power > → Creates: Certain demand for nuclear restart/new builds/SMRs > → Beneficiaries: Uranium mining → Fuel processing → Equipment manufacturing → Operators > > #### Recommended Portfolio > > | Tier | Weight | Ticker | Segment | Core Logic | > |------|------|------|------|---------| > | Core | 50% | CGN Power, Cameco | Operations + Uranium | Highest certainty | > | Satellite | 30% | China National Nuclear Power, Dongfang Electric | Operations + Equipment | Beneficiary of import substitution | > | Option | 15% | NuScale, Nano Nuclear | SMR | High risk, high beta | > | ETF | Alternative | URA, URNM | Full chain | Set-and-forget option | --- ### 5. `/industry-funnel` — Industry Funnel Screening Starting from an industry/direction: **Full market → ≤10 companies → 3 companies** layer-by-layer selection: ` Full market scan (active + top gainers + top 30 by market cap union, 30-60 names) ↓ 5 hard value investing criteria rough screen ≤10 names ↓ Fine analysis (300-500 words each) Fine analysis ≤10 names ↓ Final selection (based on portfolio complementarity, not top 3 scores) Four-master deep analysis of 3 names (800-1200 words each) ↓ Recommended portfolio (Core / Satellite / Option) + Action signals ` **Core Features**: - Each layer has clear keep/reject criteria; eliminated names leave reasons (not a black box) - Final 3 selected by "portfolio complementarity" (high certainty + medium beta + high beta), not by top scores - Mandatory column for "future IPO candidates" to avoid missing key primary market players - AI bias awareness: addresses leader bias / English bias / story bias / listing bias **Difference from `/industry-research`**: - `industry-research` focuses on industry chain structure and panorama (sliced by segments) - `industry-funnel` focuses on individual stock screening funnel (from full market down to 3 names) **Live Test: AI Sector 4 Sub-Tracks in Parallel (2026-05-09)**: | Sub-track | Final 3 | Core Position Recommendation | |-------|---------|------------| | AI Compute | TSMC / NVIDIA / SK Hynix | TSMC ★★★★★ | | AI Models | Alphabet / Meta / Alibaba | Alphabet ★★★★★ | | AI Applications | Microsoft / Adobe / AppLovin | Microsoft + Adobe ★★★★ | | AI Infrastructure Power | Eaton / TBEA / Talen Energy | Eaton + TBEA ★★★★ | **Key Insight**: The biggest winners in the AI application layer aren't AI-native companies, but established giants with distribution + data + workflow embeddedness — echoing the 1995-2000 internet bubble "picks and shovels" pattern (Amazon and Apple won, Pets.com lost). Full reports: [AI Compute](reports/AI算力-funnel-20260509.md) · [AI Models](reports/AI模型-funnel-20260509.md) · [AI Applications](reports/AI应用-funnel-20260509.md) · [AI Infrastructure Power](reports/AI基建电力-funnel-20260509.md) --- ### 6. `/private-company-research` — Deep Research on Unlisted Companies A "detective-style" research framework specifically designed for information-scarce unlisted companies: **Key Differentiators**: - **Financial Data Piecing Together**: Assemble from prospectuses, parent company reports, funding news, industry data - **Confidence Labeling**: Label each data point as 🟢High / 🟡Medium / 🔴Low confidence - **Multi-Method Valuation Cross-Verification**: Funding round valuation + comparable company analysis + DCF + terminal value back-calculation - **Exit Path Analysis**: IPO / M&A / secondary transfer — full path assessment **Output Example Excerpt**: > #### Company Snapshot: SpaceX > > | Item | Content | > |------|------| > | Latest Valuation | ~350B (2025 secondary market) 🟡 | > | Estimated Revenue | ~$13B (2024) 🟡 | > | Starlink Subscribers | 4M+ (End of 2024) 🟢 | > | Launches per Year | 100+ (2024) 🟢 | > > #### Valuation Judgment > > | Method | Valuation Range | Notes | > |——|———|——| > | Latest Funding Round | $350B | Secondary market quote, includes liquidity premium | > | Comparable Company | $200-280B | Peers: telecom + aerospace + defense | > | DCF (Baseline) | $250-350B | Assumes Starlink revenue of $30B by 2027 | > | Terminal Back-Calculation | $400-600B | Assumes Starlink becomes global telecom infrastructure | > > Blended Fair Value Range: $250B - $400B — ### 7. /news-pulse — Price Anomaly News Attribution An intelligence-response Skill designed for “quickly understanding what happened when a stock surges or plunges.” Not deep research — it’s a 10-15 minute quick attribution — to avoid panic selling or buying into rumor-driven anxiety during position anomalies. Key Differentiators: - 4-Dimensional Parallel Recon: Company events / Regulatory policy / Industry competitors / Market sentiment (sell-side + influencers + southbound flows) - Attribution Before Listing: Not just listing all news, but determining “which event justifies this price move” - Forced Nature Judgment: Value event / Sentiment-driven / True cause unknown / Mixed — “True cause unknown” is the most valuable output (potential for insider front-running) - Clear Action Recommendation: Whether to trigger deep research, review thesis, or just observe, etc. Difference from Other Skills: | Scenario | Use | |——|——| | Full research (hours) | /investment-team or /investment-research | | Deep earnings read | /earnings-review | | Long-term thesis tracking | /thesis-tracker | | Stock anomaly 10-min attribution | /news-pulse | Output Example Excerpt (Tencent 4/17-5/01 live test, 14 days -10.47%): > #### One-Sentence Attribution > This -10.47% decline is ~70-80% driven by fund flows + sentiment (buyback blackout period + southbound selling + sector beta + AI narrative taken), ~20-30% by digesting the doubled AI investment — no fundamental negative, sell-side maintains consensus Buy, nature is a “liquidity + sentiment-driven correction”, not a value event. > > #### Anomaly Attribution Table > > | Candidate Explanation | Estimated Contribution | Confidence | > |———|––––|––––| > | Buyback blackout period loss (structural, pre-Q1 results 5/13) | -3% ~ -4% | High | > | Southbound flows turning to net sell Tencent | -2% ~ -3% | High | > | AI narrative captured by competitors (DeepSeek V4/Qwen3.6/YueAn 1T) | -1% ~ -2% | Medium | > | Sector/macro beta (oil + geopolitics + Fed Warsh hawkish) | -2% ~ -3% | High | > | Pre-Q1 earnings hedging | -1% ~ -2% | Medium | > | Fundamental deterioration | 0% | Very High (excluded) | > > #### Nature Judgment: ✅ Mixed > 70% fund flow/sentiment + 20% long-term AI narrative concern + 10% pre-earnings uncertainty > > Key Counter-Evidence: Duan Yongping sold Tencent puts (bullish) on 4/8; 24 sell-side firms consensus Strong Buy; NetEase bucked the trend up 2% on 4/30 (excludes gaming industry issue); Tencent underperformed the HK Tech Index by 7 pp (HK Tech actually up 4% for the month). Invocation: /news-pulse Tencent /news-pulse Pinduoduo down12% within1week /news-pulse miHoYo — ## Live Research Reports > Below are real investment research reports generated using this framework, demonstrating the actual output quality of AI-driven investment research. | Company | Skill Used | Core Conclusion | Report Link | |——|———–|———|———| | Pinduoduo (PDD) | /investment-team | Overall 3.4/5, extremely cheap but 10-year certainty insufficient, suitable for medium position | View Report | | Tencent Holdings (0700.HK) | /investment-research | Social monopoly + excellent capital allocation; 14x forward PE reasonably low | View Report | | 7-company comparison | /investment-checklist | Moutai, Tencent Pass; NVIDIA, Meituan, Kuaishou Conditional Pass; PDD, Pop Mart Grey | View Report | | Master holdings tracking | Custom research | Latest 13F holdings of Buffett/Li Lu/Duan Yongping + PDD cost analysis | View Report | > More reports will be added. PRs welcome for reports generated using this framework. — ## Design Philosophy ### Four-Master Methodology Integration ┌──────────────────┐ │ Duan Yongping │ │ "Right Business" │ │ Business Essence │ └────────┬─────────┘ │ ┌──────────────────┼──────────────────┐ │ │ │ ▼ ▼ ▼ ┌────────┐ ┌──────────┐ ┌────────┐ │ Buffett │ │ Munger │ │ Li Lu │ │ Moat │ │ Inversion │ │ Civilization│ │ Margin of Safety │ │ Risk List │ │ Trend │ │ Management │ │ Bias Check│ │ Industry Value│ └────────┘ └──────────┘ └────────┘ The four masters aren’t simply divided labor — they’re designed to challenge each other: - Duan Yongping says “great business”, Munger asks “how could it die?” - Buffett says “cheap enough”, Li Lu asks “will it still be around in 10 years?” - What you get isn’t a concatenation of four reports, but a collision of four ways of thinking ### Financial Rigor Tools (tools/financial_rigor.py) | Feature | Command | Problem Solved | |——|——|———–| | Market Cap Verification | verify-market-cap | Price × shares precise calculation, detects unit errors | | Valuation Verification | verify-valuation | PE/PB/ROE/FCF Yield precise decimal calculation | | Multi-Source Cross-Validation | cross-validate | Auto-comparison of same data from N sources, alerts on tolerance exceedance | | Three-Scenario Valuation | three-scenario | Precise calculation of target prices for Optimistic/Baseline/Bearish | | Benford’s Law Test | benford | Detects anomalies in leading digit distribution of financial data | | Precision Calculator | calc | Precise calculation of any financial expression, replaces LLM mental math | Design Principle: All calculations use Python decimal.Decimal (precise decimal), not float (floating-point approximation). 0.1 + 0.2 = 0.3 cannot fail in financial contexts. — ## Project Roadmap - [x] Four-master comprehensive analysis framework (/investment-research) - [x] Multi-agent parallel research team (/investment-team) - [x] Buffett pre-purchase checklist (/investment-checklist) - [x] Full industry chain panorama (/industry-research + /industry-funnel) - [x] Unlisted company research framework (/private-company-research) - [x] Financial rigor tools (precise arithmetic, market cap verification, multi-source cross-validation, Benford’s Law test) - [x] Stock anomaly quick attribution (/news-pulse 4-dimension parallel recon) - [x] Deep earnings reading (/earnings-review + /earnings-team four-master parallel interpretation) - [x] Portfolio management (/portfolio-review position review and rebalancing) - [x] Investment thesis tracking (/thesis-tracker post-purchase discipline system) - [x] Management deep dive (/management-deep-dive) - [x] Quality fast screening (/quality-screen 7 hard criteria exclusion) - [x] Duan Yongping thinking simulation (/dyp-ask) - [x] Deep company series long-form article (/deep-company-series 8 articles, ~120,000 characters) - [ ] Historical backtesting: AI research vs actual stock price performance - [ ] Macroeconomic cycle analysis framework - [ ] Real-time data integration via MCP (Wind/Bloomberg/Yahoo Finance) — ## Disclaimer This project is for learning and research purposes only and does not constitute investment advice. Investing involves risk; make decisions carefully. Always do your own research (DYOR). — ## Star History If this project is helpful to you, please give it a Star! ## License MIT License — > “The best investment you can make is in yourself.” — Warren Buffett > > AI Berkshire: Giving everyone their own investment research team. Star History Chart (https://star-history.com/#xbtlin/ai-berkshire&Date)