Cross-Asset Trend Spillover: A Novel Factor for Corporate Bond Returns

Abstract

We propose XTREND, a cross-asset trend factor for corporate bond returns that captures spillovers from equity market price and volume data. Using two decades of U.S. data, we apply machine learning techniques to extract information from various technical indicators, including moving averages, oscillators, and volatility measures. The resulting signal reliably predicts corporate bond returns, demonstrating robust performance across credit quality and market conditions. Analyzing over 3.6 quadrillion models, we find that XTREND has a probability far above 50% to be part of the stochastic discount factor, suggesting its relevance for empirical bond pricing. Finally, the XTREND factor proves robust across 1.3 million research designs, expands existing bond pricing models, and reduces their pricing errors. Our findings indicate that XTREND may be driven by the frog-in-the-pan hypothesis, the anchor theory, and firm growth options.

Keywords: trend factor, corporate bonds, cross-section of bond returns, return predictability, technical analysis, cross-asset spillover, asset pricing

JEL Classification: G12, G14, G17