alpha-evaluate

Factor evaluation automates IC and ICIR computation across multiple holding periods directly from raw market data. You use Python with pandas, numpy, and scipy to process local caches or custom loaders, yielding IC time series and a comprehensive factor report.

Quintile analysis assesses long-short factor performance by dividing assets into five groups based on factor values. This Skill computes group returns to quantify monotonicity and evaluate how consistently the factor separates high and low performers.

Yes, factor evaluation requires a Python environment with pandas, numpy, and scipy installed. These libraries handle the statistical computations and data manipulation needed to generate IC metrics, quintile returns, and robustness checks.

Yes, you can use custom data loaders alongside data_cache to feed raw market data into the factor evaluation workflow. This flexibility allows you to configure the market context and compute robustness gates without being tied to a single data source.

The best way to check factor robustness is by applying robustness gates and monotonicity checks across multiple holding periods. This Skill evaluates factor quality through IC metrics and quintile analysis to determine backtest readiness and generate a factor report.

Factor backtests fail monotonicity checks when quintile returns do not consistently increase or decrease across holding periods. This Skill identifies such inconsistencies by computing ICIR time series and group returns, highlighting a lack of robust factor quality.