Technical indicators and aggregate stock returns: An updated look

Abstract

We provide updated analyses of technical indicators and aggregate stock return forecasting. We construct 105 new technical indicators as big data predictors and adopt eight advanced shrinkage methods in our forecasting analyses. Our evidence suggests that the refinements of 105 technical factors successfully overcome those of Neely et al.’s (2014) 14 technical variables to a large extent and challenge the forecasting role of Welch and Goyal's (2008) 14 popular macroeconomic variables when ENet and Lasso are used. The excellent performance of the forecasting information based on 105 technical indicators generates sufficiently high in-sample and out-of-sample R-squared values and economically sizable gains in forecasting the excess returns of the composite Standard & Poor 500 market. The corresponding evidence remains robust to changes in the business cycle, forecasting horizons, and alternative evaluation periods.