This paper introduces a novel, time-varying framework for high-dimensional factor-based asset pricing models. It leverages shrinkage techniques on regressions spanning pricing anomalies to identify statistically significant factors from a vast pool, akin to a financial Hunger Games – the Factor Games – where may the odds (p-values) be ever in their favor. The framework emphasizes sparsity, proposing methods to select a limited number of impactful factors and outperform a stricter benchmark that incorporates the Fama-French 3-factor model proposed methodology - all while avoiding look-ahead bias. Recognizing the implicit sparsity assumption in traditional models, the framework explicitly considers similar scarcity during factor selection. We apply the proposed framework to a large set of factors and various time periods, demonstrating that simple techniques can yield interesting results when applied with proper methodology. Overall, this paper provides valuable tools for researchers and practitioners, offering guidance for pricing factor selection and advocating for sparsity.
Caso aceito, me sinto confortável em apresentar o trabalho tanto em português, quanto em inglês.
Complete tool for submission, evaluation and publication of proceedings for scientific events.
