What are the key techniques for feature engineering in quant ML?
Quant ResearchLevel: Intermediate
What are the key techniques for feature engineering in quant ML?
Hints
- Identify the financial object, available information and source of risk.
- Separate theory from empirical validation with costs and constraints.
Answer
Useful features include lagged returns, rolling volatility and higher moments, cross-sectional ranks, factor interactions and appropriately timed fundamental data. Define when each input becomes available and when the resulting signal may trade. Training-derived scaling, imputation, PCA and target encoding belong inside each validation fold. Cross-sectional transforms at a date may use that date's genuinely available universe, but not future constituents or revised reports. Evaluate incremental value after costs rather than multiplying features without a hypothesis.
Source and licence: Quant Research Interview Notes — SoYuCry; English translation by photon2563 · MIT · answer: Quantoros editorial