Forecasting the sign of U.S. oil and gas industry stock index excess returns employing macroeconomic variables

In this study we propose a method of selecting the macroeconomic variables for forecasting the excess return signs of the U.S. oil and gas industry stock index by combining the Forward Sequential Variable Selection Algorithm and information criteria. We select predictors from a large monthly macroeconomic variable dataset designed by McCracken and Ng (2015). The method can adapt to the updated macroeconomic information and the possible time-varying relationship between the macroeconomic variables and the stock return signs. We also propose a method which can change the threshold value of the probit model automatically for considering the potential time-varying risk aversion level of the market participants. Further, we investigate the investment performance of an active trading strategy based on our forecasting model and compare it with a passive buy-and-hold trading strategy for different time periods.

Our study is important for both oil and gas industry investors and U.S. energy policy makers. The method that we used in this study offers a solution to the issue of selecting useful information from large datasets and absorbing updated market information.