On the directional predictability of equity premium using machine learning techniques

Abstract

This paper applies a plethora of machine learning techniques to forecast the direction of the U.S. equity premium. Our techniques include benchmark binary probit models, classification and regression trees (CART), along with penalized binary probit models. Our empirical analysis reveals that the sophisticated machine learning techniques significantly outperformed the benchmark binary probit forecasting models, both statistically and economically. Overall, the discriminant analysis classifiers are ranked first among all the models tested. Specifically, the high dimensional discriminant analysis (HDDA) classifier ranks first in terms of statistical performance, while the quadratic discriminant analysis (QDA) classifier ranks first in economic performance. The penalized likelihood binary probit models (Least Absolute Shrinkage and Selection Operator, Ridge, Elastic Net) also outperformed the benchmark binary probit models, providing significant alternatives to portfolio managers.

Forecast combinations for benchmarks of long-term stock returns using machine learning methods

“…It is often seen as " (i) a diverse collection of high-dimensional models for statistical prediction, combined with (ii) so-called 'regularization' methods for model selection and mitigation of overfit, and (iii) efficient algorithms for searching among a vast number of potential model specifications" (Gu et al, 2020). Mostly one of the following methods is well suited to address the three challenges mentioned earlier: linear models for regression (including regularization via shrinkage methods with penalization, such as Ridge Regression, Lasso, or Elastic Nets), dimension reduction via principal components regression and partial least squares, regression trees and forests (including boosted trees and random forests), (deep) neural networks, and boosting (Oztekin et al, 2016;Athey & Imbens, 2019;Coulombe et al, 2020;Gu et al, 2020;Hiabu et al, 2020;Iworiso & Vrontos, 2020;Wu et al, 2020;Gambella et al, 2021).…”

Forecasting global solar radiation using a robust regularization approach with mixture kernels

“…Zhang et al (2019) applied LASSO and EN in heterogeneous autoregressive realized volatility models to predict crude oil price. Iworiso and Vrontos (2020) proposed penalized binary probit models with LASSO and EN to forecast the direction of the US equity premium. However, neither LASSO nor EN can avoid bias estimation because they are both convex regularizations.…”

“…We make the following contributions. First, different from Iworiso and Vrontos (2020), which focused on existing regularizations, we propose a novel regularization called SCAD ridge regularization, which considers a linear combination of SCAD and ridge regularization, combining the benefits from both. Second, Huber and quantile loss functions are applied for robust forecasting.…”

Forecasting carbon price using a multi‐objective least squares support vector machine with mixture kernels

“…In the forecasting literature, several papers have attempted to tackle machine learning methods. Iworiso and Vrontos (2020) apply a plethora of machine learning techniques to forecast the direction of the US equity premium. Ntakaris et al (2018) manage the prediction of metrics in high-frequency financial markets with machine learning techniques.…”