Are the conclusions of recent studies that machine learning models can materially boost risk-adjusted stock portfolio performance reliable? In their May 2026 preliminary paper entitled "All Sizzle, No Steak! How Robust Are Financial Machine Learning Results Really?", Kristian-Alexander Janisch, Johannes Dreyer, Fuad Mehraliyev and Kristian Sund retest elastic net and neural network models as applied to stock portfolio construction in recent seminal papers with alternative samples and extended sample periods. They also investigate the incremental value of macroeconomic variables compared to firm characteristics as model inputs. Using samples of macroeconomic variables, firm characteristics, technical indicators and stock returns from prior studies, all extended through 2023, they find that:
Subscribe to Keep Reading
Get the research edge serious investors rely on.
- 1,200+ research articles
- Monthly strategy signals
- 20+ years of backtested analysis
Cancel anytime