In a current IQ-KAP research project Dr. Dominik Wolff and Dr. Fabian Echterling investigated the use of machine learning approaches for stock selection. Based on typical stock factors as well as additional fundamental data, technical indicators and historical returns, various machine learning algorithms (e.g. boosting, DNN, LSTM) are trained to predict whether a certain stock will outperform or underperform the market in the following week. Our empirical results show a substantial and significant outperformance of machine learning based stock selection models compared to a simple equally weighted benchmark. Furthermore, we show that nonlinear machine learning models such as neural networks and tree-based models outperform linear regularized logistic regression approaches. The results are robust when applied to the STOXX Europe 600 as an alternative investment universe. However, all the models analysed show significant portfolio turnover and transaction costs must be low to use the strategies profitably.