Five modules, with validation as a standing layer through every stage.
Data Engineering
Intraday data under Point-in-Time discipline — timestamps, cleaning, gaps, and multi-asset synchronization.
Market Sampling / Bar Construction
Time, volume, and dollar bars — with rigorous comparison of sampling schemes.
Feature Engineering
Price, flow, market structure, regime, cross-asset, cross-sectional, and path features.
Feature Selection
Redundancy, dependence, IC / predictive information, multicollinearity, FDR, and temporal stability.
Rule Discovery
Induction, extraction, association, subgroup, sequential mining, trees-to-rules, RuleFit, and ensembles.
Transversal Validation
Leakage → temporal integrity → multiple testing → stability → out-of-sample, in every module.