Intraday Market Return Predictability Culled from the Factor Zoo

Abstract

We provide strong empirical evidence for time-series predictability of the intraday return on the aggregate market portfolio based on lagged high-frequency cross-sectional returns from the factor zoo. Our results rely crucially on the use of modern Machine Learning techniques to regularize the predictive regressions and help tame the signals stemming from the zoo along with techniques from financial econometrics to differentiate between continuous and discontinuous price increments. Using the general prediction model for the implementation of simple trading strategies for a set of broad-based market ETFs results in sizeable out-of-sample Sharpe ratios and alphas after accounting for transaction costs. Further dissecting the results, we find that most of the superior performance may be traced to periods of high economic uncertainty and a few key factors related to tail risk and liquidity, pointing to slow-moving capital and the gradual incorporation of new information as the underlying mechanisms at work.

Keywords: High-frequency data, market return predictability, factor zoo, machine learning, market timing, market frictions, slow-moving capital

JEL Classification: G12, G14, G17, C45, C55