Missing values handling for machine learning portfolios

• 成果类型: Article
• 署名作者: Chen, Andrew Y.; McCoy, Jack
• 署名单位: Federal Reserve System - USA; Federal Reserve System Board of Governors; Columbia University
• 刊物名称: JOURNAL OF FINANCIAL ECONOMICS
• ISSN/ISSBN: 0304-405X
• DOI: 10.1016/j.jfineco.2024.103815
• 发表日期: 2024
• 关键词: Stock market predictability; Stock market anomalies; Missing values; Machine Learning

摘要

We characterize the structure and origins of missingness for 159 cross-sectional return predictors and study missing value handling for portfolios constructed using machine learning. Simply imputing with cross-sectional means performs well compared to rigorous expectation -maximization methods. This stems from three facts about predictor data: (1) missingness occurs in large blocks organized by time, (2) cross-sectional correlations are small, and (3) missingness tends to occur in blocks organized by the underlying data source. As a result, observed data provide little information about missing data. Sophisticated imputations introduce estimation noise that can lead to underperformance if machine learning is not carefully applied.