alpha-vault

name	alpha-vault
description	Portable strategy knowledge base for this repository — bundled notebooks, local markdown notes, and code implementations grouped by topic under the skill's own `ref/` folder. Use when answering research questions about these strategies, locating implementation files, or searching topic summaries from a self-contained skill bundle.

Alpha Vault

A pre-built, self-contained corpus of quant strategy references. Each topic has its own folder with a summary.md (notebook content + notes) and a source/ directory holding the original .ipynb, .md, and .py files bundled into the skill so it works after separate installation.

Prerequisites

This skill searches with rg (ripgrep). If it's not installed, install it first:

command -v rg >/dev/null || brew install ripgrep   # macOS
command -v rg >/dev/null || sudo apt-get install -y ripgrep   # Debian/Ubuntu

Workflow

1. Read the topic index to find what's relevant:

   • ref/index.md — human-readable table of all 55 topics by category
   • ref/index.json — same data, machine-readable

2. Open the matching topic's summary.md for the conceptual writeup.

3. For implementation details, read files under that topic's source/ directory (notebooks, notes, and .py files are bundled there).

4. Search across the whole corpus with rg. Always scope to ref/ to keep results focused:

   rg -n "RSRS|动量|波动率" ref/
   rg -n "NoiseArea|FScore" ref/
   rg --files-with-matches "因子择时" ref/        # just the topics that mention it
   rg -n -g '*.ipynb' "def compute" ref/          # search inside bundled notebooks
   

   Run from the skill directory so the relative ref/ path resolves; otherwise pass the absolute path to the skill's ref/ folder.

Reference Policy

• Treat the bundled summary.md as the primary conceptual source and the main ripgrep surface.
• For code questions, go to the topic's source/ files rather than re-deriving from the summary.
• The corpus is pre-built and shipped with the skill; no rebuild step is needed. Add new topics by dropping a folder under the appropriate category in ref/ with a summary.md and a source/ directory, then update ref/index.md and ref/index.json.