Scholarlink

<p>This paper introduces the Uncertainty-driven LASSO (U-LASSO), a dynamic prediction framework that adjusts the balance between long-term and short-term predictive signals based on economic uncertainty. We rationalize this adjustment through a signal extraction framework, and document that Real and Financial Uncertainty drive investors toward stable long-term signals, while Macro Uncertainty and VIX drive investors toward short-term signals. Empirical tests demonstrate that U-LASSO significantly outperforms traditional and machine learning benchmarks, in both prediction accuracy and investment performance. These findings highlight the value of integrating economic uncertainty into machine learning models to enhance both predictive performance and interpretability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

When uncertainty matters: time-varying predictability in cross-sectional stock returns

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • Shan Chen,
  • Yili Liu,
  • Qiuhua Xu

摘要

This paper introduces the Uncertainty-driven LASSO (U-LASSO), a dynamic prediction framework that adjusts the balance between long-term and short-term predictive signals based on economic uncertainty. We rationalize this adjustment through a signal extraction framework, and document that Real and Financial Uncertainty drive investors toward stable long-term signals, while Macro Uncertainty and VIX drive investors toward short-term signals. Empirical tests demonstrate that U-LASSO significantly outperforms traditional and machine learning benchmarks, in both prediction accuracy and investment performance. These findings highlight the value of integrating economic uncertainty into machine learning models to enhance both predictive performance and interpretability.