What problem does it solve?
Systematically evaluates whether cross-sectional financial factors have genuine stock-selection power by computing information coefficients (IC), information ratios (IR), and performing quantile backtests to reveal predictive strength, stability, and potential biases.
Core Features & Use Cases
- IC/IR Analysis: Compute daily IC series, summary statistics (mean, std, IR), and proportion of positive IC to judge factor direction and stability.
- Quantile Backtesting: Produce group equity curves for quantile-sorted portfolios to assess monotonicity, long-short spread, and tail effects.
- Factor Combination Methods: Support equal-weight, IC-weighted, and orthogonalized combinations for multi-factor construction and weight assignment.
- Practical Uses: Single-factor validation (momentum, value, quality), factor decay and holding-period analysis, industry-neutral screening, and multi-factor portfolio construction.
- Outputs & Requirements: Exports ic_series.csv, ic_summary.json, and group_equity.csv; requires aligned factor and forward-return CSVs (same dates and instrument columns).
Quick Start
Run the factor_analysis tool by supplying the factor CSV path, the aligned forward-return CSV path, and an output directory to generate IC series, IC summary, and quantile group equity curves.
