Return Forecasting by Quantile Regression

Abstract

A typical quantitative approach for analyzing and forecasting equity returns is to build a model based on a set of factors and then estimate the model based on a set of data and some type of least squares procedure. However, as the data in equity markets are usually far from well behaved and some standard statistical assumptions do not hold, this procedure can miss significant relationships. This article uses the quantile regression technique to reveal effects that are missed by OLS. The empirical results using S&P 500 Index data show dramatic improvement in performance using QR forecasts.

TOPICS: Factor-based models, performance measurement, factor-based models