Skip to content
Back to search

alpha-skills

Quantitative factor research skills for AI coding assistants

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/vernonoy--alpha-skills/badge.svg)](/listings/vernonoy--alpha-skills)

HTML

<a href="/listings/vernonoy--alpha-skills"><img src="/api/listings/vernonoy--alpha-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

1. Get the skills

git clone https://github.com/VernonOY/alpha-skills.git

2. Load into your AI assistant

| Platform | How | |----------|-----| | **Cursor** | Copy `skills/alpha-*/SKILL.md` → `.cursorrules` | | **Windsurf** | Copy → `.windsurfrules` | | **Claude Code** | `cp -r skills/alpha-* ~/.claude/skills/` | | **Any LLM** | Paste SKILL.md as system prompt |

3. Install Python deps

pip install pandas numpy scipy matplotlib pyarrow
pip install tushare    # A-share
pip install yfinance   # US / HK

4. Talk to it

"evaluate the momentum_20 factor"
"mine volatility factors"
"backtest my top 3 factors, 2022 to 2025"

<a id="multi-market"></a>

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 231 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <h1 align="center">Alpha Skills</h1>
  2. <p align="center">
  3. <strong>Your AI-Powered Senior Quant Researcher.</strong><br>
  4. <em>Discover alpha. Evaluate factors. Monitor decay. Backtest strategies.<br>All through natural language — in any AI coding assistant.</em>
  5. </p>
  6. <p align="center">
  7. <a href="#quick-start">Quick Start</a> · <a href="#what-can-it-do">What Can It Do</a> · <a href="#multi-market">Multi-Market</a> · <a href="#skills-reference">Skills</a> · <a href="CONTRIBUTING.md">Contribute</a>
  8. </p>
  9. ---
  10. > **Hiring a quant researcher costs $300K/year. This one is free, open-source, and works 24/7.**
  11. >
  12. > Alpha Skills turns any AI coding assistant into a senior quantitative researcher. It discovers factors, evaluates them with institutional-grade methodology (IC/ICIR/quintile/robustness), monitors for alpha decay, and runs multi-factor backtests — all from a single sentence.
  13. >
  14. > **招一个量化研究员年薪百万。这个免费、开源、7×24小时工作。**
  15. ![Alpha Skills Demo](assets/demo.png)
  16. ---
  17. ## What Can It Do
  18. **You say one sentence. It does the rest.**
  19. ```
  20. You: "Evaluate the price-volume divergence factor"
  21. AI: 📊 IC Mean=0.066 | ICIR=0.696 | Rating: ⭐ Strong
  22. Quintile spread monotonic. Best holding period: 20 days.
  23. Report saved → output/eval_pv_diverge.png
  24. You: "Mine 50 candidate factors and show me the best ones"
  25. AI: ⛏️ Scanned 50 candidates → 12 passed IC screen
  26. Top: PV divergence 20d (ICIR=0.70), Low downside vol (ICIR=0.53)...
  27. Register to library?
  28. You: "Backtest using my top 3 factors"
  29. AI: 📈 Sharpe=0.74 | MaxDD=-13.9% | Profit Factor=2.24
  30. Gate check: ✓ PF>1 ✓ MDD>-25% ✗ Sharpe<1.0
  31. ```
  32. No boilerplate. No notebooks. No 200 lines of pandas. Just results.
  33. ## Skills Reference
  34. | Skill | What It Does | Try Saying |
  35. |-------|-------------|------------|
  36. | 🔍 **alpha-discover** | Design factors from natural language | "find me a low-volatility factor" |
  37. | 📊 **alpha-evaluate** | IC / ICIR / quintile / long-short / robustness | "evaluate reversal_5" |
  38. | ⛏️ **alpha-mine** | Auto-mine factor candidates, IC screen, rank | "mine 50 factors" |
  39. | 📚 **alpha-library** | Register, list, search, retire factors (SQLite) | "show my factor library" |
  40. | 📈 **alpha-backtest** | Single & multi-factor portfolio backtest | "backtest with pv_diverge + turnover" |
  41. | 🏥 **alpha-monitor** | Detect IC decay, crowding, regime shift | "check factor health" |
  42. | 📋 **alpha-report** | Panoramic, deep-dive, comparison reports | "generate factor report" |
  43. | 📡 **alpha-signal** | Daily trading signal — target portfolio output | "today's signals" / "生成信号" |
  44. | 🤖 **alpha-autopilot** | Autonomous loop: mine → evaluate → register → monitor → retire | "run autopilot" / "自动驾驶" |
  45. <a id="quick-start"></a>
  46. ## Quick Start
  47. ### 1. Get the skills
  48. ```bash
  49. git clone https://github.com/VernonOY/alpha-skills.git
  50. ```
  51. ### 2. Load into your AI assistant
  52. | Platform | How |
  53. |----------|-----|
  54. | **Cursor** | Copy `skills/alpha-*/SKILL.md` → `.cursorrules` |
  55. | **Windsurf** | Copy → `.windsurfrules` |
  56. | **Claude Code** | `cp -r skills/alpha-* ~/.claude/skills/` |
  57. | **Any LLM** | Paste SKILL.md as system prompt |
  58. ### 3. Install Python deps
  59. ```bash
  60. pip install pandas numpy scipy matplotlib pyarrow
  61. pip install tushare # A-share
  62. pip install yfinance # US / HK
  63. ```
  64. ### 4. Talk to it
  65. ```
  66. "evaluate the momentum_20 factor"
  67. "mine volatility factors"
  68. "backtest my top 3 factors, 2022 to 2025"
  69. ```
  70. <a id="multi-market"></a>
  71. ## Multi-Market: A-Share · Hong Kong · US
  72. Works out of the box for three markets. Auto-adapts trading rules per market:
  73. | | A-share 🇨🇳 | Hong Kong 🇭🇰 | US 🇺🇸 |
  74. |---|---|---|---|
  75. | **Data** | Tushare Pro | Yahoo Finance | Yahoo Finance |
  76. | **Price Limit** | ±10% | None | None |
  77. | **T+N** | T+1 | T+0 | T+0 |
  78. | **Cost** | 0.3% | 0.2% | 0.1% |
  79. | **Benchmark** | CSI 300 | HSI | S&P 500 |
  80. | **Pool** | 5000+ stocks | 78 HSI constituents | 143 S&P 500 |
  81. Switch markets in one line:
  82. ```markdown
  83. MARKET: US
  84. DATA_MODULE: examples.us_data_yfinance
  85. ```
  86. **Bring your own data.** Write a 7-function Python adapter for Bloomberg, AkShare, Binance, or any source — [see interface spec](examples/README.md).
  87. ## How It Works
  88. ```
  89. ┌─────────────────────────────────────────────┐
  90. │ You (natural language) │
  91. ├─────────────────────────────────────────────┤
  92. │ AI Coding Assistant │
  93. │ (Cursor / Windsurf / Claude Code / ...) │
  94. ├─────────────────────────────────────────────┤
  95. │ Alpha Skills (7 SKILL.md) │
  96. │ discover · evaluate · mine · library │
  97. │ backtest · monitor · report │
  98. ├─────────────────────────────────────────────┤
  99. │ Python (pandas/numpy/scipy/matplotlib) │
  100. │ → factor computation │
  101. │ → IC/ICIR/quintile evaluation │
  102. │ → portfolio backtesting │
  103. │ → SQLite factor registry │
  104. ├─────────────────────────────────────────────┤
  105. │ Data: Tushare │ YFinance │ CSV │ Custom │
  106. └─────────────────────────────────────────────┘
  107. ```
  108. **Zero framework dependency.** Each skill is a self-contained Markdown file. The AI reads it, writes the Python, runs it. Nothing to install except standard data science packages.
  109. ## Evaluation Pipeline
  110. Your AI quant researcher doesn't just compute IC. It runs a **4-level institutional-grade evaluation**:
  111. | Level | What | Speed |
  112. |-------|------|-------|
  113. | **L0** | Syntax + data validation | instant |
  114. | **L1** | Quick IC screen (sampled 200 stocks × 2 years) | <30s |
  115. | **L2** | Full: IC series, ICIR, quintile returns, long-short, monotonicity | 1-3 min |
  116. | **L3** | Robustness: parameter perturbation, rolling window, start-date sensitivity | 5-15 min |
  117. Plus optional **[qtype](https://github.com/VernonOY/qtype)** pre-flight — static analysis to catch look-ahead bias before you waste compute on fake alpha.
  118. ## Factor Mining Engine
  119. `alpha-mine` systematically searches the factor expression space:
  120. **3 mining strategies:**
  121. - **Template-based** — momentum, mean-reversion, volatility, volume, composite templates × multiple window sizes
  122. - **Combinatorial** — chain operators: `cs_rank(ts_corr(close, volume, 20))`
  123. - **Mutation** — take a known strong factor, mutate parameters/operators
  124. **Pipeline:** Generate 50+ candidates → IC quick screen → full evaluate top 10 → LLM judges economic intuition → present ranked results
  125. **Overfitting guard:** Every surviving factor gets an **economic intuition score** (Strong / Moderate / Weak). Factors without a clear behavioral story are flagged as potential data mining.
  126. ## Built-in Factors (25+)
  127. | Category | Factors |
  128. |----------|---------|
  129. | **Price-Volume** | momentum · reversal · volatility · pv_diverge · rsi · macd · bollinger · atr_ratio · turnover · abnormal_turnover |
  130. | **Fundamental** | roe · roa · gross_margin · net_profit_growth · revenue_growth |
  131. | **Valuation** | pe_ttm · pb · ps_ttm · dividend_yield · peg |
  132. | **Composite** | quality_score · value_score · growth_momentum |
  133. All gate checks and evaluation thresholds are **user-configurable**:
  134. ```markdown
  135. GATE_SHARPE: 1.0
  136. GATE_MAX_DRAWDOWN: -0.25
  137. GATE_PROFIT_FACTOR: 1.0
  138. EVAL_ICIR_STRONG: 0.5
  139. ```
  140. ## What's New
  141. **v0.3 — Autopilot & Live Signals**
  142. - `alpha-signal`: daily trading signal generator — outputs target portfolio from active factors
  143. - `alpha-autopilot`: autonomous research loop — auto-mine, evaluate, register, monitor, retire
  144. - Professional knowledge base: 6 expert-level reference documents (2,795 lines)
  145. **v0.2 — Automated Factor Mining**
  146. - `alpha-mine`: systematically search factor expression space, IC screen, economic intuition scoring
  147. - All skills fully self-contained — zero external package dependencies
  148. - Optional [qtype](https://github.com/VernonOY/qtype) pre-flight check
  149. **v0.1 — Initial Release**
  150. - 7 core skills · A-share/HK/US support · bilingual EN/ZH · multi-platform
  151. ## Roadmap
  152. - [x] 9 skills (discover / evaluate / mine / library / backtest / monitor / report / signal / autopilot)
  153. - [x] Daily signal generation (target portfolio output)
  154. - [x] Autonomous research loop (mine → evaluate → register → monitor → retire)
  155. - [x] Professional knowledge base (6 expert-level documents, 2,795 lines)
  156. - [x] A-share, HK, and US market support
  157. - [x] Market-aware trading rules
  158. - [x] Automated factor mining (template + combinatorial + mutation)
  159. - [x] Custom data source support
  160. - [x] Multi-platform (Cursor, Windsurf, Claude Code, ChatGPT, local models)
  161. - [x] [qtype](https://github.com/VernonOY/qtype) integration for static code checks
  162. - [ ] Portfolio construction (factor → tradeable portfolio)
  163. - [ ] Market regime detection & factor-regime mapping
  164. - [ ] Factor crowding detection
  165. - [ ] Web UI dashboard
  166. ## License
  167. Apache 2.0
  168. ## Contributing
  169. [See CONTRIBUTING.md](CONTRIBUTING.md) — add skills, data adapters, or improve methodology.
  170. ---
  171. <p align="center">
  172. <strong>Stop writing boilerplate. Start finding alpha.</strong><br>
  173. <sub>Built by quants who got tired of copy-pasting the same IC calculation for the 500th time.</sub>
  174. </p>

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.