Practical Applications of Harvesting Multi-Asset Carry, Value, and Momentum: Work Smarter, Not Harder

Abstract

In <b><i>Harvesting Multi-Asset Carry, Value, and Momentum: Work Smarter, Not Harder</i></b>, from the Spring 2022 issue of <b><i>The Journal of Financial Data Science</i></b>, authors <b>Brian Jacobsen</b> and <b>Matthias Scheiber</b> (both of <b>Allspring Global Investments</b>) investigate the effectiveness of different systematic investing strategies. The traditional approach treats carry, value, and momentum as trading signals that dictate when to establish a position. The study finds that these signals strengthen or decay, depending on how long the investor waits to make the initial trade or holds the assets thereafter. Investment returns vary widely and are marginally negative, on average. The study then tests out a different approach that uses carry, value, and momentum not as triggers for trades but as explanatory variables in a machine learning–based decision-tree model that determines which assets are sending the strongest signals. This approach produces marginally positive returns when it compares each asset’s signals to the median of its asset class, and it produces significantly positive returns when it compares each asset’s signals to the median for all asset classes. Returns improve even more when the investor frequently monitors trades and signals and closes out positions when signals decay.