Skip to content
Get packs

Factor Research Framework

Test whether stock factors predict returns using IC/IR stats and quantile backtests

Source Download skill
Repository
HKUDS/Vibe-Trading
Stars
37,433
Licence
MIT
Files
4 files

What it does

When installed, the agent can evaluate whether a numeric signal (such as momentum, value, or volatility) actually helps pick winning instruments across a group of stocks or other assets. It guides the agent to build aligned factor and forward-return tables, call a factor_analysis tool, and judge results using IC/IR statistics and quantile group backtests. It can also help combine several factors into a composite using equal, IC-based, or orthogonalized weights.

Great for

  • Quantitative researchers screening candidate alpha factors across many instruments
  • Analysts who want a structured, look-ahead-aware way to compare factors and weight them

Skip it if

  • Casual investors looking for stock picks: the skill evaluates factor statistics, not investment advice, and backtest results can mislead when overfit
  • Single-asset price forecasting: the framework is designed for cross-sectional evaluation across multiple instruments

Try asking

  • “Test whether a 20-day momentum factor has predictive power across my stock universe”
  • “Compare IC and IR for value and quality factors and tell me which to keep”
  • “Build an IC-weighted composite of my three factors using a 252-day trailing window”
  • “Check whether my volatility factor shows monotonic quantile group returns”
Step-by-step guide: how to use Factor Research FrameworkInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 4 original files, the MIT licence and an attribution note.

    factor-research.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip factor-research.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip factor-research.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Verify that factor and return CSVs align exactly by date and code, and that returns are forward returns after the observation date, to avoid look-ahead bias
  • Treat IC mean above 0.10 as a warning sign to check for data leakage rather than a success
  • Use out-of-sample testing before relying on any factor or weighting scheme

What's inside

  • references/combination.md
  • references/neutralization.md
  • references/research_contract.md
  • SKILL.md

Redistributed unmodified from HKUDS/Vibe-Trading under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.