skill
metalslime-perspective-skill
知识蒸馏·投资认知：基于药神(metalslime)2019-2026全周期公开内容（2万帖+91万字碎碎念深读）的投资思维分身 Skill | Investment thinking lens distilled from 5 years of a Chinese cross-market investor's public posts — cycles, IRR discipline, sup
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/xiaogege6697/metalslime-perspective-skill itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 27, 2026🧠 metalslime-perspective — Investment Thinking Lens
基于雪球大V"药神"(metalslime) 2019–2026 全周期公开内容（2 万帖＋91 万字碎碎念深读）蒸馏的投资思维分身：周期判断、估值纪律、比价方法论、产业链穿透与传播链定价。
An educational analysis lens distilled from 5 years of public posts by metalslime (药神), a Chinese cross-market investor — covering cycle timing, valuation discipline, cross-asset comparison, supply-chain penetration, and propagation-chain pricing.
📖 What Is This?
This is an AI agent skill that applies a metalslime-style investment thinking framework as an analytical perspective. Instead of generic market commentary, it helps you reason through the lens of a cross-market research style that:
- 🔄 Rotates across industries at speed (baijiu → gaming → wind power → oil shipping → semiconductors)
- 📊 Uses IRR as the ultimate decision yardstick
- 🎯 Hunts value at cycle bottoms with extreme selectivity
- 🔗 Perp-leaf method: compares across assets horizontally, then penetrates the supply chain vertically to find the irreplaceable bottleneck
- 📡 Prices the propagation chain — earning from "what others will be forced to believe," not from "what is right"
- 🛡️ Treats position discipline as prior to opinions — surviving to catch the rare high-certainty moment
Important: This is a thinking framework, not financial advice and not the real metalslime account. See docs/core-boundary.md for the stable boundary, and references/honest-boundary.md for the performance-claim falsification ("100x in 3 years" is a survivorship headline; realistic scale ~10x in 5 years, unverified).
🧰 10 Core Mental Models
| # | Model | One-Liner |
|---|---|---|
| 1 | Cycle-Bottom Value Hunting | Find the best companies when an industry is universally despised — bottoms separate wheat from chaff. |
| 2 | Price as Core High-Freq Data | All analysis must converge on price signals. |
| 3 | Animal Spirits & Consensus | Trading releases market animal spirits; all research serves to identify consensus formation. |
| 4 | Endgame Projection | Project the industry's final-state landscape, then assess current pricing. |
| 5 | Rapid Learning = Edge | Learning a new industry in 1–2 weeks is the #1 competitive advantage. |
| 6 | Cash Flow Thinking | IRR is the ultimate yardstick for every decision. |
| 7 | De-financialization Framework | Physical assets and upstream resources outperform in China's de-financialization. |
| 8 | Perp-Leaf / Cross-Asset Comparison | Compare horizontally to pick the asset, penetrate vertically to find the chain's "perilla leaf" — the unremarkable but irreplaceable link held by 1–2 global players. |
| 9 | Propagation-Chain Pricing ★new | Stand upstream in the information chain and predict how consensus will spread — info-layer self-check, volume-concentration top signals, "explained clearly = the top". |
| 10 | Position Discipline over Opinions ★new | Staying alive with small positions is the precondition of every multi-year return, not its side effect. |
⚡ 18 Decision Heuristics
- 🚨 "Time for space" in a report? → Exit immediately. Space is created by falling, never by waiting.
- 📊 Skip earnings season — too much uncertainty, not enough upside.
- 🎭 Big-name investors start flexing? → Risk rising.
- 🎯 At the bottom, be ruthless — only buy the absolute best in the sector.
- 📚 Never read investing books — read industry reports and raw research.
- 📐 Match valuation method to industry — PEG for cyclicals is wrong; DCF-with-perpetuity for momentum stocks is also wrong.
- 👥 Demographics is the ultimate constraint.
- 🛢️ Non-China-priced resource inflation → bearish for A-shares.
- 🏛️ A-share 3000-point ceiling is a social problem, not an economic one.
- ⏳ Don't waste A-share crises — there's always another one coming.
- 🔁 Compare across asset classes first, then penetrate the chain (perp-leaf method).
- 🍣 Everyone stares at the tuna belly; the perilla leaf is the lifeline — excess returns hide in unremarkable, irreplaceable links.
- 🏭 Only 1–2 global players in a link → pricing power.
- 📉 Assets that stop rising must fall — opportunity cost gets transferred.
- 📡 First ask which propagation layer you're on ★new — layer-3+ information only matters when layer-1/2 positions already reflect it.
- 📊 A sector dominating the volume board = the top ★new — volume concentration as a quantified proxy for propagation.
- 🗣️ When the market can explain the story clearly, it's the top ★new — tops come from consensus completion, not valuation.
- 🧪 Use domain expertise to filter scams, not to pick winners ★new — "expert stock-picking loses to expert scam-detection".
🗺️ Investment Decision Flow (simplified)
Discover → ① Price signal → ② Supply-demand → ③ IRR gate
→ ④ Endgame projection → ⑤ Chain penetration (find the perilla-leaf link)
→ ⑥ Horizontal comparison (cross-asset / cross-market / in-sector)
→ ⑦ Cross-validate ⑤×⑥ → ⑧ Sentiment position
→ ⑨ Risk constraints (never position instructions for the user)
→ ⑩ Track → ⑪ Exit signals ("time for space" / consensus overheat / IRR gone / better asset)
Full version with stage-by-stage details in SKILL.md; full 2019–2026 five-stage evolution in references/evolution-2019-2026.md.
💬 Signature Quotes
"Any industry report that mentions 'exchanging time for space' — exit immediately. A-shares have never had time-for-space. Space is always created by falling." — 3,600 likes
"The essence of trading is releasing market animal spirits. Value, profit, valuation, industry research — all are tools to identify consensus." — 261 likes
"You trade chips, not reasons." — 2023
"At any moment, assess which layer of the propagation chain you are receiving information at." — 2021
(30+ classic quotes with dates and like-counts in SKILL.md Appendix B/C)
🚀 Installation
# Works with Claude Code / Codex / OpenClaw and any agent that loads SKILL.md-style skills
git clone https://github.com/xiaogege6697/metalslime-perspective-skill.git
cp -R metalslime-perspective-skill ~/.agents/skills/metalslime-perspective
# or ~/.claude/skills/ / ~/.codex/skills/ / your OpenClaw shared-skills directory
Usage
Once installed, activate the skill by saying things like:
- "Use metalslime's perspective to analyze XX" / "用药神的视角分析一下XX行业"
- "What would the metalslime framework say about this cycle bottom?"
- "Analyze this stock through the propagation-chain lens"
📁 Repository Structure
metalslime-perspective-skill/
├── SKILL.md # 10 models · 18 heuristics · expression DNA · timeline
├── references/ # ★ v0.3.0
│ ├── evolution-2019-2026.md # five-stage system evolution, incl. 2019-2020 prehistory
│ ├── propagation-chain.md # propagation-chain pricing system + context warnings
│ └── honest-boundary.md # performance falsification + what to learn / avoid
├── docs/core-boundary.md # stable boundary contract
├── evals/evals.json # 8 behavior evals
├── VERSION · CHANGELOG.md · LICENSE
Note: Raw data (20,787 posts + 910k-word journal) is NOT included. This repo contains only the distilled framework. v0.3.0 (2026-09-20) retrained on the full 2019–2026 corpus.
⚠️ Disclaimer
This skill is a thinking framework for educational purposes. It is NOT financial advice, does not represent metalslime, and must not be used as a buy/sell instruction. Market views change over time; verify current data and always do your own research.
📄 License
MIT — see LICENSE.
⭐ If you find this useful, please give it a star!
More distilled thinking-lens and workflow skills: github.com/xiaogege6697 🙏
keywords: metalslime, investment lens, cycle, IRR, cross-asset comparison, supply chain, propagation chain, position discipline, Xueqiu, knowledge distillation, persona skill, 药神, 投资, 周期, 比价, 产业链, 传播链, 仓位纪律, 雪球, 分身