Predictability of future stock returns by combination of predictors

Abstract

This dissertation evaluates the return predictability of forecasts derived from Fama-MacBeth regressions. I show how investors can combine firm characteristics to estimate the subsequent month’s stock returns in cross section. These forecasts exhibit a high predictive ability as they capture the cross-sectional dispersion of returns consistently and are independent of size, bookto- market ratio and industry. Combining a set of nine firm characteristics, the return estimates exhibit a standard deviation of 1% and yield a predictive slope of 0.85 for large stocks. Portfolio sorts based on these forecasts offer a Sharpe ratio of 1.06 for a simple buy and hold strategy and a Sharpe Ratio of 1.24 for a long-short investment strategy net of transaction costs.

Date of Award	26 Oct 2016
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	José Faias (Supervisor)