Subsampled factor models for asset pricing: The rise of Vasa

We propose a new method, variable subsample aggregation (VASA), for equity return prediction using large-dimensional set of factors. To demonstrate the effectiveness, robustness, and dimension reduction power VASA, we perform comparative analysis between state-of-the-art machine learning algorithms. As performance measure, explore not only global predictive but also stock-specific R 2 's their distribution. While reflects average forecasting accuracy, find that high variability in can be detrimental portfolio performance. Since VASA shows minimal variability, portfolios formed on this method outperform based random forests neural nets.