The factor zoo challenges financial research and raised doubts regarding post-publication robustness and data snooping. Feng et al. (2020) find that only a few new anomalies are enough to explain the cross section of stock returns, while the other factors are redundant. Another way of handling the factor zoo are unsupervised machine learning techniques that target dimensionality reduction. Instead of direct competition between the factors, manifold learning techniques allow mapping factors to a lower dimensional setting and as such concentrate patterns into few exploitable signals.