Investing with Momentum: The Past, Present, and Future

Abstract

Stock selection models often use momentum, analysts’ expectations, and fundamental data. We found support for composite modeling using these sources of data for U.S. equities during the 1998–2007 period. We found additional evidence to support the use of Barra and APT multifactor models for portfolio construction and risk control. Three levels of testing of stock selection and portfolio construction models were developed and estimated. We created portfolios for the January 1998–December 2007 period. We report three conclusions: 1) Momentum investing was rewarded by the market in the United States from December 1928–December 2007; 2) Momentum can be combined with reported fundamental data—such as earnings, book value, cash flow, and sales—and analysts’ earnings forecast revisions in a stock selection model to identify mispriced securities; 3) the portfolio returns of the multifactor risk-controlled portfolio returns allow us to reject the data mining corrections test null hypothesis.

TOPICS: Portfolio construction, factor-based models, big data/machine learning